Krzysztof Czarnecki

Prarthana Bhattacharyya, Chengjie Huang, Krzysztof Czarnecki

Self-supervised learning (SSL) is an emerging technique that has been successfully employed to train convolutional neural networks (CNNs) and graph neural networks (GNNs) for more transferable, generalizable, and robust representation learning. However its potential in motion forecasting for autonomous driving has rarely been explored. In this study, we report the first systematic exploration and assessment of incorporating self-supervision into motion forecasting. We first propose to investigate four novel self-supervised learning tasks for motion forecasting with theoretical rationale and quantitative and qualitative comparisons on the challenging large-scale Argoverse dataset. Secondly, we point out that our auxiliary SSL-based learning setup not only outperforms forecasting methods which use transformers, complicated fusion mechanisms and sophisticated online dense goal candidate optimization algorithms in terms of performance accuracy, but also has low inference time and architectural complexity. Lastly, we conduct several experiments to understand why SSL improves motion forecasting. Code is open-sourced at \url{https://github.com/AutoVision-cloud/SSL-Lanes}.

Matthew Pitropov, Chengjie Huang, Vahdat Abdelzad et al. · utoronto

The estimation of uncertainty in robotic vision, such as 3D object detection, is an essential component in developing safe autonomous systems aware of their own performance. However, the deployment of current uncertainty estimation methods in 3D object detection remains challenging due to timing and computational constraints. To tackle this issue, we propose LiDAR-MIMO, an adaptation of the multi-input multi-output (MIMO) uncertainty estimation method to the LiDAR-based 3D object detection task. Our method modifies the original MIMO by performing multi-input at the feature level to ensure the detection, uncertainty estimation, and runtime performance benefits are retained despite the limited capacity of the underlying detector and the large computational costs of point cloud processing. We compare LiDAR-MIMO with MC dropout and ensembles as baselines and show comparable uncertainty estimation results with only a small number of output heads. Further, LiDAR-MIMO can be configured to be twice as fast as MC dropout and ensembles, while achieving higher mAP than MC dropout and approaching that of ensembles.

Luke Rowe, Martin Ethier, Eli-Henry Dykhne et al.

Predicting the future motion of road agents is a critical task in an autonomous driving pipeline. In this work, we address the problem of generating a set of scene-level, or joint, future trajectory predictions in multi-agent driving scenarios. To this end, we propose FJMP, a Factorized Joint Motion Prediction framework for multi-agent interactive driving scenarios. FJMP models the future scene interaction dynamics as a sparse directed interaction graph, where edges denote explicit interactions between agents. We then prune the graph into a directed acyclic graph (DAG) and decompose the joint prediction task into a sequence of marginal and conditional predictions according to the partial ordering of the DAG, where joint future trajectories are decoded using a directed acyclic graph neural network (DAGNN). We conduct experiments on the INTERACTION and Argoverse 2 datasets and demonstrate that FJMP produces more accurate and scene-consistent joint trajectory predictions than non-factorized approaches, especially on the most interactive and kinematically interesting agents. FJMP ranks 1st on the multi-agent test leaderboard of the INTERACTION dataset.

Chengjie Huang, Van Duong Nguyen, Vahdat Abdelzad et al.

3D object detection is an essential part of automated driving, and deep neural networks (DNNs) have achieved state-of-the-art performance for this task. However, deep models are notorious for assigning high confidence scores to out-of-distribution (OOD) inputs, that is, inputs that are not drawn from the training distribution. Detecting OOD inputs is challenging and essential for the safe deployment of models. OOD detection has been studied extensively for the classification task, but it has not received enough attention for the object detection task, specifically LiDAR-based 3D object detection. In this paper, we focus on the detection of OOD inputs for LiDAR-based 3D object detection. We formulate what OOD inputs mean for object detection and propose to adapt several OOD detection methods for object detection. We accomplish this by our proposed feature extraction method. To evaluate OOD detection methods, we develop a simple but effective technique of generating OOD objects for a given object detection model. Our evaluation based on the KITTI dataset shows that different OOD detection methods have biases toward detecting specific OOD objects. It emphasizes the importance of combined OOD detection methods and more research in this direction.

Scott Larter, Rodrigo Queiroz, Sean Sedwards et al.

Modelling pedestrian behavior is crucial in the development and testing of autonomous vehicles. In this work, we present a hierarchical pedestrian behavior model that generates high-level decisions through the use of behavior trees, in order to produce maneuvers executed by a low-level motion planner using an adapted Social Force model. A full implementation of our work is integrated into GeoScenario Server, a scenario definition and execution engine, extending its vehicle simulation capabilities with pedestrian simulation. The extended environment allows simulating test scenarios involving both vehicles and pedestrians to assist in the scenario-based testing process of autonomous vehicles. The presented hierarchical model is evaluated on two real-world data sets collected at separate locations with different road structures. Our model is shown to replicate the real-world pedestrians' trajectories with a high degree of fidelity and a decision-making accuracy of 98% or better, given only high-level routing information for each pedestrian.

Atrisha Sarkar, Kate Larson, Krzysztof Czarnecki

A central design problem in game theoretic analysis is the estimation of the players' utilities. In many real-world interactive situations of human decision making, including human driving, the utilities are multi-objective in nature; therefore, estimating the parameters of aggregation, i.e., mapping of multi-objective utilities to a scalar value, becomes an essential part of game construction. However, estimating this parameter from observational data introduces several challenges due to a host of unobservable factors, including the underlying modality of aggregation and the possibly boundedly rational behaviour model that generated the observation. Based on the concept of rationalisability, we develop algorithms for estimating multi-objective aggregation parameters for two common aggregation methods, weighted and satisficing aggregation, and for both strategic and non-strategic reasoning models. Based on three different datasets, we provide insights into how human drivers aggregate the utilities of safety and progress, as well as the situational dependence of the aggregation process. Additionally, we show that irrespective of the specific solution concept used for solving the games, a data-driven estimation of utility aggregation significantly improves the predictive accuracy of behaviour models with respect to observed human behaviour.

Sunsheng Gu, Vahdat Abdelzad, Krzysztof Czarnecki

Explainable AI (XAI) methods are frequently applied to obtain qualitative insights about deep models' predictions. However, such insights need to be interpreted by a human observer to be useful. In this paper, we aim to use explanations directly to make decisions without human observers. We adopt two gradient-based explanation methods, Integrated Gradients (IG) and backprop, for the task of 3D object detection. Then, we propose a set of quantitative measures, named Explanation Concentration (XC) scores, that can be used for downstream tasks. These scores quantify the concentration of attributions within the boundaries of detected objects. We evaluate the effectiveness of XC scores via the task of distinguishing true positive (TP) and false positive (FP) detected objects in the KITTI and Waymo datasets. The results demonstrate an improvement of more than 100\% on both datasets compared to other heuristics such as random guesses and the number of LiDAR points in the bounding box, raising confidence in XC's potential for application in more use cases. Our results also indicate that computationally expensive XAI methods like IG may not be more valuable when used quantitatively compare to simpler methods.

Rick Salay, Krzysztof Czarnecki

Although artificial intelligence-based perception (AIP) using deep neural networks (DNN) has achieved near human level performance, its well-known limitations are obstacles to the safety assurance needed in autonomous applications. These include vulnerability to adversarial inputs, inability to handle novel inputs and non-interpretability. While research in addressing these limitations is active, in this paper, we argue that a fundamentally different approach is needed to address them. Inspired by dual process models of human cognition, where Type 1 thinking is fast and non-conscious while Type 2 thinking is slow and based on conscious reasoning, we propose a dual process architecture for safe AIP. We review research on how humans address the simplest non-trivial perception problem, image classification, and sketch a corresponding AIP architecture for this task. We argue that this architecture can provide a systematic way of addressing the limitations of AIP using DNNs and an approach to assurance of human-level performance and beyond. We conclude by discussing what components of the architecture may already be addressed by existing work and what remains future work.

Luke Rowe, Benjamin Thérien, Krzysztof Czarnecki et al.

In adversarial machine learning, the popular $\ell_\infty$ threat model has been the focus of much previous work. While this mathematical definition of imperceptibility successfully captures an infinite set of additive image transformations that a model should be robust to, this is only a subset of all transformations which leave the semantic label of an image unchanged. Indeed, previous work also considered robustness to spatial attacks as well as other semantic transformations; however, designing defense methods against the composition of spatial and $\ell_{\infty}$ perturbations remains relatively underexplored. In the following, we improve the understanding of this seldom investigated compositional setting. We prove theoretically that no linear classifier can achieve more than trivial accuracy against a composite adversary in a simple statistical setting, illustrating its difficulty. We then investigate how state-of-the-art $\ell_{\infty}$ defenses can be adapted to this novel threat model and study their performance against compositional attacks. We find that our newly proposed TRADES$_{\text{All}}$ strategy performs the strongest of all. Analyzing its logit's Lipschitz constant for RT transformations of different sizes, we find that TRADES$_{\text{All}}$ remains stable over a wide range of RT transformations with and without $\ell_\infty$ perturbations.

Benjamin Thérien, Krzysztof Czarnecki

Imagine experiencing a crash as the passenger of an autonomous vehicle. Wouldn't you want to know why it happened? Current end-to-end optimizable deep neural networks (DNNs) in 3D detection, multi-object tracking, and motion forecasting provide little to no explanations about how they make their decisions. To help bridge this gap, we design an end-to-end optimizable multi-object tracking architecture and training protocol inspired by the recently proposed method of interchange intervention training (IIT). By enumerating different tracking decisions and associated reasoning procedures, we can train individual networks to reason about the possible decisions via IIT. Each network's decisions can be explained by the high-level structural causal model (SCM) it is trained in alignment with. Moreover, our proposed model learns to rank these outcomes, leveraging the promise of deep learning in end-to-end training, while being inherently interpretable.

Mozhgan Nasr Azadani, Yimu Wang, Yongpeng Zhu et al.

Establishing a clear link between model predictions and the visual evidence that supports them is critical for transparency and reliability in multimodal reasoning, yet current multimodal large language model (MLLM) evaluations do not explicitly enforce this alignment. Existing benchmarks assess either textual answer correctness or pixel-level localization in isolation, leaving the coupling of reasoning and grounding an open challenge. We introduce VISTAQA, a comprehensive benchmark for joint evaluation of free-form answer correctness and pixel-level evidence grounding in visual question answering. VISTAQA comprises 1,157 expert-curated samples spanning six task types and six visual domains, ranging from direct perception to compositional and relational reasoning. VISTAQA requires models to not only answer correctly, but to also provide precise segmentation masks that support their answers. It also includes hallucination-aware examples where no valid visual evidence exists. To support this enhanced evaluation, we introduce GROVE, a unified evaluation metric that enforces joint correctness by combining textual accuracy and grounding quality via a per-sample geometric mean, ensuring neither dimension can compensate for deficiencies in the other. Comprehensive experiments across grounding-aware models and hybrid pipelines with general-purpose MLLMs reveal that even the strongest systems achieve limited performance under GROVE, highlighting a substantial gap between answer accuracy and visual evidence alignment.

Mozhgan Nasr Azadani, James Riddell, Sean Sedwards et al.

Enhanced visual understanding serves as a cornerstone for multimodal large language models (MLLMs). Recent hybrid MLLMs incorporate a mixture of vision experts to address the limitations of using a single vision encoder and excessively long visual tokens. Despite the progress of these MLLMs, a research gap remains in effectively integrating diverse vision encoders. This work explores fusion strategies of visual tokens for hybrid MLLMs, leading to the design of LEO, a novel MLLM with a dual-branch vision encoder framework that incorporates a post-adaptation fusion strategy and adaptive tiling: for each segmented tile of the input images, LEO sequentially interleaves the visual tokens from its two vision encoders. Extensive evaluation across 13 vision-language benchmarks reveals that LEO outperforms state-of-the-art open-source MLLMs and hybrid MLLMs on the majority of tasks. Furthermore, we show that LEO can be adapted to the specialized domain of autonomous driving without altering the model architecture or training recipe, achieving competitive performance compared to existing baselines. The code and model will be publicly available.

Yimu Wang, Mozhgan Nasr Azadani, Sean Sedwards et al.

Improving the visual understanding ability of vision-language models (VLMs) is crucial for enhancing their performance across various tasks. While using multiple pretrained visual experts has shown great promise, it often incurs significant computational costs during training and inference. To address this challenge, we propose HAWAII, a novel framework that distills knowledge from multiple visual experts into a single vision encoder, enabling it to inherit the complementary strengths of several experts with minimal computational overhead. To mitigate conflicts among different teachers and switch between different teacher-specific knowledge, instead of using a fixed set of adapters for multiple teachers, we propose to use teacher-specific Low-Rank Adaptation (LoRA) adapters with a corresponding router. Each adapter is aligned with a specific teacher, avoiding noisy guidance during distillation. To enable efficient knowledge distillation, we propose fine-grained and coarse-grained distillation. At the fine-grained level, token importance scores are employed to emphasize the most informative tokens from each teacher adaptively. At the coarse-grained level, we summarize the knowledge from multiple teachers and transfer it to the student using a set of general-knowledge LoRA adapters with a router. Extensive experiments on various vision-language tasks demonstrate the superiority of HAWAII compared to popular open-source VLMs. The code is available at https://github.com/yimuwangcs/wise-hawaii.

Prarthana Bhattacharyya, Chengjie Huang, Krzysztof Czarnecki

Existing point-cloud based 3D object detectors use convolution-like operators to process information in a local neighbourhood with fixed-weight kernels and aggregate global context hierarchically. However, non-local neural networks and self-attention for 2D vision have shown that explicitly modeling long-range interactions can lead to more robust and competitive models. In this paper, we propose two variants of self-attention for contextual modeling in 3D object detection by augmenting convolutional features with self-attention features. We first incorporate the pairwise self-attention mechanism into the current state-of-the-art BEV, voxel and point-based detectors and show consistent improvement over strong baseline models of up to 1.5 3D AP while simultaneously reducing their parameter footprint and computational cost by 15-80% and 30-50%, respectively, on the KITTI validation set. We next propose a self-attention variant that samples a subset of the most representative features by learning deformations over randomly sampled locations. This not only allows us to scale explicit global contextual modeling to larger point-clouds, but also leads to more discriminative and informative feature descriptors. Our method can be flexibly applied to most state-of-the-art detectors with increased accuracy and parameter and compute efficiency. We show our proposed method improves 3D object detection performance on KITTI, nuScenes and Waymo Open datasets. Code is available at https://github.com/AutoVision-cloud/SA-Det3D.

Erkan Baser, Venkateshwaran Balasubramanian, Prarthana Bhattacharyya et al.

We propose a data-driven approach to online multi-object tracking (MOT) that uses a convolutional neural network (CNN) for data association in a tracking-by-detection framework. The problem of multi-target tracking aims to assign noisy detections to a-priori unknown and time-varying number of tracked objects across a sequence of frames. A majority of the existing solutions focus on either tediously designing cost functions or formulating the task of data association as a complex optimization problem that can be solved effectively. Instead, we exploit the power of deep learning to formulate the data association problem as inference in a CNN. To this end, we propose to learn a similarity function that combines cues from both image and spatial features of objects. Our solution learns to perform global assignments in 3D purely from data, handles noisy detections and a varying number of targets, and is easy to train. We evaluate our approach on the challenging KITTI dataset and show competitive results. Our code is available at https://git.uwaterloo.ca/wise-lab/fantrack.

Matt Angus, Mohamed ElBalkini, Samin Khan et al.

In training deep neural networks for semantic segmentation, the main limiting factor is the low amount of ground truth annotation data that is available in currently existing datasets. The limited availability of such data is due to the time cost and human effort required to accurately and consistently label real images on a pixel level. Modern sandbox video game engines provide open world environments where traffic and pedestrians behave in a pseudo-realistic manner. This caters well to the collection of a believable road-scene dataset. Utilizing open-source tools and resources found in single-player modding communities, we provide a method for persistent, ground truth, asset annotation of a game world. By collecting a synthetic dataset containing upwards of $1,000,000$ images, we demonstrate real-time, on-demand, ground truth data annotation capability of our method. Supplementing this synthetic data to Cityscapes dataset, we show that our data generation method provides qualitative as well as quantitative improvements---for training networks---over previous methods that use video games as surrogate.

Chengjie Huang, Vahdat Abdelzad, Sean Sedwards et al.

We consider the problem of cross-sensor domain adaptation in the context of LiDAR-based 3D object detection and propose Stationary Object Aggregation Pseudo-labelling (SOAP) to generate high quality pseudo-labels for stationary objects. In contrast to the current state-of-the-art in-domain practice of aggregating just a few input scans, SOAP aggregates entire sequences of point clouds at the input level to reduce the sensor domain gap. Then, by means of what we call quasi-stationary training and spatial consistency post-processing, the SOAP model generates accurate pseudo-labels for stationary objects, closing a minimum of 30.3% domain gap compared to few-frame detectors. Our results also show that state-of-the-art domain adaptation approaches can achieve even greater performance in combination with SOAP, in both the unsupervised and semi-supervised settings.

Jie Wang, Yash Vardhan Pant, Lei Zhao et al.

With the increasing presence of autonomous vehicles (AVs) on public roads, developing robust control strategies to navigate the uncertainty of human-driven vehicles (HVs) is crucial. This paper introduces an advanced method for modeling HV behavior, combining a first-principles model with Gaussian process (GP) learning to enhance velocity prediction accuracy and provide a measurable uncertainty. We validated this innovative HV model using real-world data from field experiments and applied it to develop a GP-enhanced model predictive control (GP-MPC) strategy. This strategy aims to improve safety in mixed vehicle platoons by integrating uncertainty assessment into distance constraints. Comparative simulation studies with a conventional model predictive control (MPC) approach demonstrated that our GP-MPC strategy ensures more reliable safe distancing and fosters efficient vehicular dynamics, achieving notably higher speeds within the platoon. By incorporating a sparse GP technique in HV modeling and adopting a dynamic GP prediction within the MPC framework, we significantly reduced the computation time of GP-MPC, marking it only 4.6% higher than that of the conventional MPC. This represents a substantial improvement, making the process about 100 times faster than our preliminary work without these approximations. Our findings underscore the effectiveness of learning-based HV modeling in enhancing both safety and operational efficiency in mixed-traffic environments, paving the way for more harmonious AV-HV interactions.

Yimu Wang, Evelien Riddell, Adrian Chow et al.

Existing vision-language model (VLM)-based methods for out-of-distribution (OOD) detection typically rely on similarity scores between input images and in-distribution (ID) text prototypes. However, the modality gap between image and text often results in high false positive rates, as OOD samples can exhibit high similarity to ID text prototypes. To mitigate the impact of this modality gap, we propose incorporating ID image prototypes along with ID text prototypes. We present theoretical analysis and empirical evidence indicating that this approach enhances VLM-based OOD detection performance without any additional training. To further reduce the gap between image and text, we introduce a novel few-shot tuning framework, SUPREME, comprising biased prompts generation (BPG) and image-text consistency (ITC) modules. BPG enhances image-text fusion and improves generalization by conditioning ID text prototypes on the Gaussian-based estimated image domain bias; ITC reduces the modality gap by minimizing intra- and inter-modal distances. Moreover, inspired by our theoretical and empirical findings, we introduce a novel OOD score $S_{\textit{GMP}}$, leveraging uni- and cross-modal similarities. Finally, we present extensive experiments to demonstrate that SUPREME consistently outperforms existing VLM-based OOD detection methods.

Prarthana Bhattacharyya, Chengjie Huang, Krzysztof Czarnecki

This paper addresses motion forecasting in multi-agent environments, pivotal for ensuring safety of autonomous vehicles. Traditional as well as recent data-driven marginal trajectory prediction methods struggle to properly learn non-linear agent-to-agent interactions. We present SSL-Interactions that proposes pretext tasks to enhance interaction modeling for trajectory prediction. We introduce four interaction-aware pretext tasks to encapsulate various aspects of agent interactions: range gap prediction, closest distance prediction, direction of movement prediction, and type of interaction prediction. We further propose an approach to curate interaction-heavy scenarios from datasets. This curated data has two advantages: it provides a stronger learning signal to the interaction model, and facilitates generation of pseudo-labels for interaction-centric pretext tasks. We also propose three new metrics specifically designed to evaluate predictions in interactive scenes. Our empirical evaluations indicate SSL-Interactions outperforms state-of-the-art motion forecasting methods quantitatively with up to 8% improvement, and qualitatively, for interaction-heavy scenarios.

Yimu Wang, Mozhgan Nasr Azadani, Sean Sedwards et al.

Redundancy of visual tokens in multi-modal large language models (MLLMs) significantly reduces their computational efficiency. Recent approaches, such as resamplers and summarizers, have sought to reduce the number of visual tokens, but at the cost of visual reasoning ability. To address this, we propose LEO-MINI, a novel MLLM that significantly reduces the number of visual tokens and simultaneously boosts visual reasoning capabilities. For efficiency, LEO-MINI incorporates CoTR, a novel token reduction module to consolidate a large number of visual tokens into a smaller set of tokens, using the similarity between visual tokens, text tokens, and a compact learnable query. For effectiveness, to scale up the model's ability with minimal computational overhead, LEO-MINI employs MMoE, a novel mixture of multi-modal experts module. MMOE employs a set of LoRA experts with a novel router to switch between them based on the input text and visual tokens instead of only using the input hidden state. MMoE also includes a general LoRA expert that is always activated to learn general knowledge for LLM reasoning. For extracting richer visual features, MMOE employs a set of vision experts trained on diverse domain-specific data. To demonstrate LEO-MINI's improved efficiency and performance, we evaluate it against existing efficient MLLMs on various benchmark vision-language tasks.

Huakun Shen, Boyue Caroline Hu, Krzysztof Czarnecki et al.

While Neural Networks (NNs) have surpassed human accuracy in image classification on ImageNet, they often lack robustness against image corruption, i.e., corruption robustness. Yet such robustness is seemingly effortless for human perception. In this paper, we propose visually-continuous corruption robustness (VCR) -- an extension of corruption robustness to allow assessing it over the wide and continuous range of changes that correspond to the human perceptive quality (i.e., from the original image to the full distortion of all perceived visual information), along with two novel human-aware metrics for NN evaluation. To compare VCR of NNs with human perception, we conducted extensive experiments on 14 commonly used image corruptions with 7,718 human participants and state-of-the-art robust NN models with different training objectives (e.g., standard, adversarial, corruption robustness), different architectures (e.g., convolution NNs, vision transformers), and different amounts of training data augmentation. Our study showed that: 1) assessing robustness against continuous corruption can reveal insufficient robustness undetected by existing benchmarks; as a result, 2) the gap between NN and human robustness is larger than previously known; and finally, 3) some image corruptions have a similar impact on human perception, offering opportunities for more cost-effective robustness assessments. Our validation set with 14 image corruptions, human robustness data, and the evaluation code is provided as a toolbox and a benchmark.

Mei Qi Tang, Sean Sedwards, Chengjie Huang et al.

The impact of snowfall on 3D object detection performance remains underexplored. Conducting such an evaluation requires a dataset with sufficient labelled data from both weather conditions, ideally captured in the same driving environment. Current driving datasets with LiDAR point clouds either do not provide enough labelled data in both snowy and clear weather conditions, or rely on de-snowing methods to generate synthetic clear weather. Synthetic data often lacks realism and introduces an additional domain shift that confounds accurate evaluations. To address these challenges, we present CADC+, the first paired weather domain adaptation dataset for autonomous driving in winter conditions. CADC+ extends the Canadian Adverse Driving Conditions dataset (CADC) using clear weather data that was recorded on the same roads and in the same period as CADC. To create CADC+, we pair each CADC sequence with a clear weather sequence that matches the snowy sequence as closely as possible. CADC+ thus minimizes the domain shift resulting from factors unrelated to the presence of snow. We also present some preliminary results using CADC+ to evaluate the effect of snow on 3D object detection performance. We observe that snow introduces a combination of aleatoric and epistemic uncertainties, acting as both noise and a distinct data domain.

Chengjie Huang, Krzysztof Czarnecki

We propose MFSeg, an efficient multi-frame 3D semantic segmentation framework. By aggregating point cloud sequences at the feature level and regularizing the feature extraction and aggregation process, MFSeg reduces computational overhead while maintaining high accuracy. Moreover, by employing a lightweight MLP-based point decoder, our method eliminates the need to upsample redundant points from past frames. Experiments on the nuScenes and Waymo datasets show that MFSeg outperforms existing methods, demonstrating its effectiveness and efficiency.

Adrian Chow, Evelien Riddell, Yimu Wang et al.

Open-vocabulary 3D object detection for autonomous driving aims to detect novel objects beyond the predefined training label sets in point cloud scenes. Existing approaches achieve this by connecting traditional 3D object detectors with vision-language models (VLMs) to regress 3D bounding boxes for novel objects and perform open-vocabulary classification through cross-modal alignment between 3D and 2D features. However, achieving robust cross-modal alignment remains a challenge due to semantic inconsistencies when generating corresponding 3D and 2D feature pairs. To overcome this challenge, we present OV-SCAN, an Open-Vocabulary 3D framework that enforces Semantically Consistent Alignment for Novel object discovery. OV-SCAN employs two core strategies: discovering precise 3D annotations and filtering out low-quality or corrupted alignment pairs (arising from 3D annotation, occlusion-induced, or resolution-induced noise). Extensive experiments on the nuScenes dataset demonstrate that OV-SCAN achieves state-of-the-art performance.

Chengjie Huang, Vahdat Abdelzad, Sean Sedwards et al.

Input aggregation is a simple technique used by state-of-the-art LiDAR 3D object detectors to improve detection. However, increasing aggregation is known to have diminishing returns and even performance degradation, due to objects responding differently to the number of aggregated frames. To address this limitation, we propose an efficient adaptive method, which we call Variable Aggregation Detection (VADet). Instead of aggregating the entire scene using a fixed number of frames, VADet performs aggregation per object, with the number of frames determined by an object's observed properties, such as speed and point density. VADet thus reduces the inherent trade-offs of fixed aggregation and is not architecture specific. To demonstrate its benefits, we apply VADet to three popular single-stage detectors and achieve state-of-the-art performance on the Waymo dataset.

Krzysztof Czarnecki, Hiroshi Kuwajima

Driving Automation Systems (DAS) are subject to complex road environments and vehicle behaviors and increasingly rely on sophisticated sensors and Artificial Intelligence (AI). These properties give rise to unique safety faults stemming from specification insufficiencies and technological performance limitations, where sensors and AI introduce errors that vary in magnitude and temporal patterns, posing potential safety risks. The Safety of the Intended Functionality (SOTIF) standard emerges as a promising framework for addressing these concerns, focusing on scenario-based analysis to identify hazardous behaviors and their causes. Although the current standard provides a basic cause-and-effect model and high-level process guidance, it lacks concepts required to identify and evaluate hazardous errors, especially within the context of AI. This paper introduces two key contributions to bridge this gap. First, it defines the SOTIF Temporal Error and Failure Model (STEAM) as a refinement of the SOTIF cause-and-effect model, offering a comprehensive system-design perspective. STEAM refines error definitions, introduces error sequences, and classifies them as error sequence patterns, providing particular relevance to systems employing advanced sensors and AI. Second, this paper proposes the Model-based SOTIF Analysis of Failures and Errors (MoSAFE) method, which allows instantiating STEAM based on system-design models by deriving hazardous error sequence patterns at module level from hazardous behaviors at vehicle level via weakest precondition reasoning. Finally, the paper presents a case study centered on an automated speed-control feature, illustrating the practical applicability of the refined model and the MoSAFE method in addressing complex safety challenges in DAS.

Benjamin Thérien, Chengjie Huang, Adrian Chow et al.

Object re-identification (ReID) from images plays a critical role in application domains of image retrieval (surveillance, retail analytics, etc.) and multi-object tracking (autonomous driving, robotics, etc.). However, systems that additionally or exclusively perceive the world from depth sensors are becoming more commonplace without any corresponding methods for object ReID. In this work, we fill the gap by providing the first large-scale study of object ReID from point clouds and establishing its performance relative to image ReID. To enable such a study, we create two large-scale ReID datasets with paired image and LiDAR observations and propose a lightweight matching head that can be concatenated to any set or sequence processing backbone (e.g., PointNet or ViT), creating a family of comparable object ReID networks for both modalities. Run in Siamese style, our proposed point cloud ReID networks can make thousands of pairwise comparisons in real-time ($10$ Hz). Our findings demonstrate that their performance increases with higher sensor resolution and approaches that of image ReID when observations are sufficiently dense. Our strongest network trained at the largest scale achieves ReID accuracy exceeding $90\%$ for rigid objects and $85\%$ for deformable objects (without any explicit skeleton normalization). To our knowledge, we are the first to study object re-identification from real point cloud observations.

Boyue Caroline Hu, Lina Marsso, Krzysztof Czarnecki et al.

Machine Vision Components (MVC) are becoming safety-critical. Assuring their quality, including safety, is essential for their successful deployment. Assurance relies on the availability of precisely specified and, ideally, machine-verifiable requirements. MVCs with state-of-the-art performance rely on machine learning (ML) and training data but largely lack such requirements. In this paper, we address the need for defining machine-verifiable reliability requirements for MVCs against transformations that simulate the full range of realistic and safety-critical changes in the environment. Using human performance as a baseline, we define reliability requirements as: 'if the changes in an image do not affect a human's decision, neither should they affect the MVC's.' To this end, we provide: (1) a class of safety-related image transformations; (2) reliability requirement classes to specify correctness-preservation and prediction-preservation for MVCs; (3) a method to instantiate machine-verifiable requirements from these requirements classes using human performance experiment data; (4) human performance experiment data for image recognition involving eight commonly used transformations, from about 2000 human participants; and (5) a method for automatically checking whether an MVC satisfies our requirements. Further, we show that our reliability requirements are feasible and reusable by evaluating our methods on 13 state-of-the-art pre-trained image classification models. Finally, we demonstrate that our approach detects reliability gaps in MVCs that other existing methods are unable to detect.

Jaeyoung Lee, Sean Sedwards, Krzysztof Czarnecki

We consider the challenge of finding a deterministic policy for a Markov decision process that uniformly (in all states) maximizes one reward subject to a probabilistic constraint over a different reward. Existing solutions do not fully address our precise problem definition, which nevertheless arises naturally in the context of safety-critical robotic systems. This class of problem is known to be hard, but the combined requirements of determinism and uniform optimality can create learning instability. In this work, after describing and motivating our problem with a simple example, we present a suitable constrained reinforcement learning algorithm that prevents learning instability, using recursive constraints. Our proposed approach admits an approximative form that improves efficiency and is conservative w.r.t. the constraint.

Atrisha Sarkar, Kate Larson, Krzysztof Czarnecki

In order to enable autonomous vehicles (AV) to navigate busy traffic situations, in recent years there has been a focus on game-theoretic models for strategic behavior planning in AVs. However, a lack of common taxonomy impedes a broader understanding of the strategies the models generate as well as the development of safety specification to identity what strategies are safe for an AV to execute. Based on common patterns of interaction in traffic conflicts, we develop a taxonomy for strategic interactions along the dimensions of agents' initial response to right-of-way rules and subsequent response to other agents' behavior. Furthermore, we demonstrate a process of automatic mapping of strategies generated by a strategic planner to the categories in the taxonomy, and based on vehicle-vehicle and vehicle-pedestrian interaction simulation, we evaluate two popular solution concepts used in strategic planning in AVs, QLk and Subgame perfect $ε$-Nash Equilibrium, with respect to those categories.

Atrisha Sarkar, Kate Larson, Krzysztof Czarnecki

While there has been an increasing focus on the use of game theoretic models for autonomous driving, empirical evidence shows that there are still open questions around dealing with the challenges of common knowledge assumptions as well as modeling bounded rationality. To address some of these practical challenges, we develop a framework of generalized dynamic cognitive hierarchy for both modelling naturalistic human driving behavior as well as behavior planning for autonomous vehicles (AV). This framework is built upon a rich model of level-0 behavior through the use of automata strategies, an interpretable notion of bounded rationality through safety and maneuver satisficing, and a robust response for planning. Based on evaluation on two large naturalistic datasets as well as simulation of critical traffic scenarios, we show that i) automata strategies are well suited for level-0 behavior in a dynamic level-k framework, and ii) the proposed robust response to a heterogeneous population of strategic and non-strategic reasoners can be an effective approach for game theoretic planning in AV.

Maximilian Kahn, Atrisha Sarkar, Krzysztof Czarnecki

A particular challenge for both autonomous and human driving is dealing with risk associated with dynamic occlusion, i.e., occlusion caused by other vehicles in traffic. Based on the theory of hypergames, we develop a novel multi-agent dynamic occlusion risk (DOR) measure for assessing situational risk in dynamic occlusion scenarios. Furthermore, we present a white-box, scenario-based, accelerated safety validation framework for assessing safety of strategic planners in AV. Based on evaluation over a large naturalistic database, our proposed validation method achieves a 4000% speedup compared to direct validation on naturalistic data, a more diverse coverage, and ability to generalize beyond the dataset and generate commonly observed dynamic occlusion crashes in traffic in an automated manner.

Rick Salay, Krzysztof Czarnecki, Hiroshi Kuwajima et al.

Safety assurance is a central concern for the development and societal acceptance of automated driving (AD) systems. Perception is a key aspect of AD that relies heavily on Machine Learning (ML). Despite the known challenges with the safety assurance of ML-based components, proposals have recently emerged for unit-level safety cases addressing these components. Unfortunately, AD safety cases express safety requirements at the system level and these efforts are missing the critical linking argument needed to integrate safety requirements at the system level with component performance requirements at the unit level. In this paper, we propose the Integration Safety Case for Perception (ISCaP), a generic template for such a linking safety argument specifically tailored for perception components. The template takes a deductive and formal approach to define strong traceability between levels. We demonstrate the applicability of ISCaP with a detailed case study and discuss its use as a tool to support incremental development of perception components.

Tsutomu Kobayashi, Rick Salay, Ichiro Hasuo et al.

Formal reasoning on the safety of controller systems interacting with plants is complex because developers need to specify behavior while taking into account perceptual uncertainty. To address this, we propose an automated workflow that takes an Event-B model of an uncertainty-unaware controller and a specification of uncertainty as input. First, our workflow automatically injects the uncertainty into the original model to obtain an uncertainty-aware but potentially unsafe controller. Then, it automatically robustifies the controller so that it satisfies safety even under the uncertainty. The case study shows how our workflow helps developers to explore multiple levels of perceptual uncertainty. We conclude that our workflow makes design and analysis of uncertainty-aware controller systems easier and more systematic.

Henry Chen, Robin Cohen, Kerstin Dautenhahn et al.

Autonomous Vehicles (AV) will transform transportation, but also the interaction between vehicles and pedestrians. In the absence of a driver, it is not clear how an AV can communicate its intention to pedestrians. One option is to use visual signals. To advance their design, we conduct four human-participant experiments and evaluate six representative AV visual signals for visibility, intuitiveness, persuasiveness, and usability at pedestrian crossings. Based on the results, we distill twelve practical design recommendations for AV visual signals, with focus on signal pattern design and placement. Moreover, the paper advances the methodology for experimental evaluation of visual signals, including lab, closed-course, and public road tests using an autonomous vehicle. In addition, the paper also reports insights on pedestrian crosswalk behaviours and the impacts of pedestrian trust towards AVs on the behaviors. We hope that this work will constitute valuable input to the ongoing development of international standards for AV lamps, and thus help mature automated driving in general.

Atrisha Sarkar, Krzysztof Czarnecki

With autonomous vehicles (AV) set to integrate further into regular human traffic, there is an increasing consensus on treating AV motion planning as a multi-agent problem. However, the traditional game-theoretic assumption of complete rationality is too strong for human driving, and there is a need for understanding human driving as a \emph{bounded rational} activity through a behavioural game-theoretic lens. To that end, we adapt four metamodels of bounded rational behaviour: three based on Quantal level-k and one based on Nash equilibrium with quantal errors. We formalize the different solution concepts that can be applied in the context of hierarchical games, a framework used in multi-agent motion planning, for the purpose of creating game theoretic models of driving behaviour. Furthermore, based on a contributed dataset of human driving at a busy urban intersection with a total of approximately 4k agents and 44k decision points, we evaluate the behaviour models on the basis of model fit to naturalistic data, as well as their predictive capacity. Our results suggest that among the behaviour models evaluated, at the level of maneuvers, modeling driving behaviour as an adaptation of the Quantal level-k model with level-0 behaviour modelled as pure rule-following provides the best fit to naturalistic driving behaviour. At the level of trajectories, bounds sampling of actions and a maxmax non-strategic models is the most accurate within the set of models in comparison. We also find a significant impact of situational factors on the performance of behaviour models.

Prarthana Bhattacharyya, Krzysztof Czarnecki

We present Deformable PV-RCNN, a high-performing point-cloud based 3D object detector. Currently, the proposal refinement methods used by the state-of-the-art two-stage detectors cannot adequately accommodate differing object scales, varying point-cloud density, part-deformation and clutter. We present a proposal refinement module inspired by 2D deformable convolution networks that can adaptively gather instance-specific features from locations where informative content exists. We also propose a simple context gating mechanism which allows the keypoints to select relevant context information for the refinement stage. We show state-of-the-art results on the KITTI dataset.

Vahdat Abdelzad, Krzysztof Czarnecki, Rick Salay

Deep neural networks (DNNs) have become the de facto learning mechanism in different domains. Their tendency to perform unreliably on out-of-distribution (OOD) inputs hinders their adoption in critical domains. Several approaches have been proposed for detecting OOD inputs. However, existing approaches still lack robustness. In this paper, we shed light on the robustness of OOD detection (OODD) approaches by revealing the important role of optimization methods. We show that OODD approaches are sensitive to the type of optimization method used during training deep models. Optimization methods can provide different solutions to a non-convex problem and so these solutions may or may not satisfy the assumptions (e.g., distributions of deep features) made by OODD approaches. Furthermore, we propose a robustness score that takes into account the role of optimization methods. This provides a sound way to compare OODD approaches. In addition to comparing several OODD approaches using our proposed robustness score, we demonstrate that some optimization methods provide better solutions for OODD approaches.

Ryan De Iaco, Stephen L. Smith, Krzysztof Czarnecki

This paper characterizes safe following distances for on-road driving when vehicles can avoid collisions by either braking or by swerving into an adjacent lane. In particular, we focus on safety as defined in the Responsibility-Sensitive Safety (RSS) framework. We extend RSS by introducing swerve manoeuvres as a valid response in addition to the already present brake manoeuvre. These swerve manoeuvres use the more realistic kinematic bicycle model rather than the double integrator model of RSS. When vehicles are able to swerve and brake, it is shown that their required safe following distance at higher speeds is less than that required through braking alone. In addition, when all vehicles follow this new distance, they are provably safe. The use of the kinematic bicycle model is then validated by comparing these swerve manoeuvres to that of a dynamic single-track model.

Matthew Pitropov, Danson Garcia, Jason Rebello et al.

The Canadian Adverse Driving Conditions (CADC) dataset was collected with the Autonomoose autonomous vehicle platform, based on a modified Lincoln MKZ. The dataset, collected during winter within the Region of Waterloo, Canada, is the first autonomous vehicle dataset that focuses on adverse driving conditions specifically. It contains 7,000 frames collected through a variety of winter weather conditions of annotated data from 8 cameras (Ximea MQ013CG-E2), Lidar (VLP-32C) and a GNSS+INS system (Novatel OEM638). The sensors are time synchronized and calibrated with the intrinsic and extrinsic calibrations included in the dataset. Lidar frame annotations that represent ground truth for 3D object detection and tracking have been provided by Scale AI.

Matt Angus, Krzysztof Czarnecki, Rick Salay

The detection of out of distribution samples for image classification has been widely researched. Safety critical applications, such as autonomous driving, would benefit from the ability to localise the unusual objects causing the image to be out of distribution. This paper adapts state-of-the-art methods for detecting out of distribution images for image classification to the new task of detecting out of distribution pixels, which can localise the unusual objects. It further experimentally compares the adapted methods on two new datasets derived from existing semantic segmentation datasets using PSPNet and DeeplabV3+ architectures, as well as proposing a new metric for the task. The evaluation shows that the performance ranking of the compared methods does not transfer to the new task and every method performs significantly worse than their image-level counterparts.

Vahdat Abdelzad, Krzysztof Czarnecki, Rick Salay et al.

Deep neural networks achieve superior performance in challenging tasks such as image classification. However, deep classifiers tend to incorrectly classify out-of-distribution (OOD) inputs, which are inputs that do not belong to the classifier training distribution. Several approaches have been proposed to detect OOD inputs, but the detection task is still an ongoing challenge. In this paper, we propose a new OOD detection approach that can be easily applied to an existing classifier and does not need to have access to OOD samples. The detector is a one-class classifier trained on the output of an early layer of the original classifier fed with its original training set. We apply our approach to several low- and high-dimensional datasets and compare it to the state-of-the-art detection approaches. Our approach achieves substantially better results over multiple metrics.

Sachin Vernekar, Ashish Gaurav, Vahdat Abdelzad et al.

By design, discriminatively trained neural network classifiers produce reliable predictions only for in-distribution samples. For their real-world deployments, detecting out-of-distribution (OOD) samples is essential. Assuming OOD to be outside the closed boundary of in-distribution, typical neural classifiers do not contain the knowledge of this boundary for OOD detection during inference. There have been recent approaches to instill this knowledge in classifiers by explicitly training the classifier with OOD samples close to the in-distribution boundary. However, these generated samples fail to cover the entire in-distribution boundary effectively, thereby resulting in a sub-optimal OOD detector. In this paper, we analyze the feasibility of such approaches by investigating the complexity of producing such "effective" OOD samples. We also propose a novel algorithm to generate such samples using a manifold learning network (e.g., variational autoencoder) and then train an n+1 classifier for OOD detection, where the $n+1^{th}$ class represents the OOD samples. We compare our approach against several recent classifier-based OOD detectors on MNIST and Fashion-MNIST datasets. Overall the proposed approach consistently performs better than the others.

Braden Hurl, Robin Cohen, Krzysztof Czarnecki et al.

Inter-vehicle communication for autonomous vehicles (AVs) stands to provide significant benefits in terms of perception robustness. We propose a novel approach for AVs to communicate perceptual observations, tempered by trust modelling of peers providing reports. Based on the accuracy of reported object detections as verified locally, communicated messages can be fused to augment perception performance beyond line of sight and at great distance from the ego vehicle. Also presented is a new synthetic dataset which can be used to test cooperative perception. The TruPercept dataset includes unreliable and malicious behaviour scenarios to experiment with some challenges cooperative perception introduces. The TruPercept runtime and evaluation framework allows modular component replacement to facilitate ablation studies as well as the creation of new trust scenarios we are able to show.

Jian Deng, Krzysztof Czarnecki

This paper presents Multi-view Labelling Object Detector (MLOD). The detector takes an RGB image and a LIDAR point cloud as input and follows the two-stage object detection framework. A Region Proposal Network (RPN) generates 3D proposals in a Bird's Eye View (BEV) projection of the point cloud. The second stage projects the 3D proposal bounding boxes to the image and BEV feature maps and sends the corresponding map crops to a detection header for classification and bounding-box regression. Unlike other multi-view based methods, the cropped image features are not directly fed to the detection header, but masked by the depth information to filter out parts outside 3D bounding boxes. The fusion of image and BEV features is challenging, as they are derived from different perspectives. We introduce a novel detection header, which provides detection results not just from fusion layer, but also from each sensor channel. Hence the object detector can be trained on data labelled in different views to avoid the degeneration of feature extractors. MLOD achieves state-of-the-art performance on the KITTI 3D object detection benchmark. Most importantly, the evaluation shows that the new header architecture is effective in preventing image feature extractor degeneration.

Marko Ilievski, Sean Sedwards, Ashish Gaurav et al.

We explore the complex design space of behaviour planning for autonomous driving. Design choices that successfully address one aspect of behaviour planning can critically constrain others. To aid the design process, in this work we decompose the design space with respect to important choices arising from the current state of the art approaches, and describe the resulting trade-offs. In doing this, we also identify interesting directions of future work.

Changjian Li, Krzysztof Czarnecki

The standard reinforcement learning (RL) formulation considers the expectation of the (discounted) cumulative reward. This is limiting in applications where we are concerned with not only the expected performance, but also the distribution of the performance. In this paper, we introduce micro-objective reinforcement learning --- an alternative RL formalism that overcomes this issue. In this new formulation, a RL task is specified by a set of micro-objectives, which are constructs that specify the desirability or undesirability of events. In addition, micro-objectives allow prior knowledge in the form of temporal abstraction to be incorporated into the global RL objective. The generality of this formalism, and its relations to single/multi-objective RL, and hierarchical RL are discussed.

Braden Hurl, Krzysztof Czarnecki, Steven Waslander

We introduce the Precise Synthetic Image and LiDAR (PreSIL) dataset for autonomous vehicle perception. Grand Theft Auto V (GTA V), a commercial video game, has a large detailed world with realistic graphics, which provides a diverse data collection environment. Existing works creating synthetic LiDAR data for autonomous driving with GTA V have not released their datasets, rely on an in-game raycasting function which represents people as cylinders, and can fail to capture vehicles past 30 metres. Our work creates a precise LiDAR simulator within GTA V which collides with detailed models for all entities no matter the type or position. The PreSIL dataset consists of over 50,000 frames and includes high-definition images with full resolution depth information, semantic segmentation (images), point-wise segmentation (point clouds), and detailed annotations for all vehicles and people. Collecting additional data with our framework is entirely automatic and requires no human annotation of any kind. We demonstrate the effectiveness of our dataset by showing an improvement of up to 5% average precision on the KITTI 3D Object Detection benchmark challenge when state-of-the-art 3D object detection networks are pre-trained with our data. The data and code are available at https://tinyurl.com/y3tb9sxy

Sachin Vernekar, Ashish Gaurav, Taylor Denouden et al.

Discriminatively trained neural classifiers can be trusted, only when the input data comes from the training distribution (in-distribution). Therefore, detecting out-of-distribution (OOD) samples is very important to avoid classification errors. In the context of OOD detection for image classification, one of the recent approaches proposes training a classifier called "confident-classifier" by minimizing the standard cross-entropy loss on in-distribution samples and minimizing the KL divergence between the predictive distribution of OOD samples in the low-density regions of in-distribution and the uniform distribution (maximizing the entropy of the outputs). Thus, the samples could be detected as OOD if they have low confidence or high entropy. In this paper, we analyze this setting both theoretically and experimentally. We conclude that the resulting confident-classifier still yields arbitrarily high confidence for OOD samples far away from the in-distribution. We instead suggest training a classifier by adding an explicit "reject" class for OOD samples.