Jen Jen Chung

Chunwei Xing, Xinyu Sun, Andrei Cramariuc et al. · eth-zurich

Current descriptors for global localization often struggle under vast viewpoint or appearance changes. One possible improvement is the addition of topological information on semantic objects. However, handcrafted topological descriptors are hard to tune and not robust to environmental noise, drastic perspective changes, object occlusion or misdetections. To solve this problem, we formulate a learning-based approach by modelling semantically meaningful object constellations as graphs and using Deep Graph Convolution Networks to map a constellation to a descriptor. We demonstrate the effectiveness of our Deep Learned Constellation Descriptor (Descriptellation) on two real-world datasets. Although Descriptellation is trained on randomly generated simulation datasets, it shows good generalization abilities on real-world datasets. Descriptellation also outperforms state-of-the-art and handcrafted constellation descriptors for global localization, and is robust to different types of noise. The code is publicly available at https://github.com/ethz-asl/Descriptellation.

Daniel Dugas, Olov Andersson, Roland Siegwart et al.

Autonomously navigating a robot in everyday crowded spaces requires solving complex perception and planning challenges. When using only monocular image sensor data as input, classical two-dimensional planning approaches cannot be used. While images present a significant challenge when it comes to perception and planning, they also allow capturing potentially important details, such as complex geometry, body movement, and other visual cues. In order to successfully solve the navigation task from only images, algorithms must be able to model the scene and its dynamics using only this channel of information. We investigate whether the world model concept, which has shown state-of-the-art results for modeling and learning policies in Atari games as well as promising results in 2D LiDAR-based crowd navigation, can also be applied to the camera-based navigation problem. To this end, we create simulated environments where a robot must navigate past static and moving humans without colliding in order to reach its goal. We find that state-of-the-art methods are able to achieve success in solving the navigation problem, and can generate dream-like predictions of future image-sequences which show consistent geometry and moving persons. We are also able to show that policy performance in our high-fidelity sim2real simulation scenario transfers to the real world by testing the policy on a real robot. We make our simulator, models and experiments available at https://github.com/danieldugas/NavDreams.

Kenneth Blomqvist, Francesco Milano, Jen Jen Chung et al.

Recently, groundbreaking results have been presented on open-vocabulary semantic image segmentation. Such methods segment each pixel in an image into arbitrary categories provided at run-time in the form of text prompts, as opposed to a fixed set of classes defined at training time. In this work, we present a zero-shot volumetric open-vocabulary semantic scene segmentation method. Our method builds on the insight that we can fuse image features from a vision-language model into a neural implicit representation. We show that the resulting feature field can be segmented into different classes by assigning points to natural language text prompts. The implicit volumetric representation enables us to segment the scene both in 3D and 2D by rendering feature maps from any given viewpoint of the scene. We show that our method works on noisy real-world data and can run in real-time on live sensor data dynamically adjusting to text prompts. We also present quantitative comparisons on the ScanNet dataset.

Kenneth Blomqvist, Lionel Ott, Jen Jen Chung et al.

Methods have recently been proposed that densely segment 3D volumes into classes using only color images and expert supervision in the form of sparse semantically annotated pixels. While impressive, these methods still require a relatively large amount of supervision and segmenting an object can take several minutes in practice. Such systems typically only optimize their representation on the particular scene they are fitting, without leveraging any prior information from previously seen images. In this paper, we propose to use features extracted with models trained on large existing datasets to improve segmentation performance. We bake this feature representation into a Neural Radiance Field (NeRF) by volumetrically rendering feature maps and supervising on features extracted from each input image. We show that by baking this representation into the NeRF, we make the subsequent classification task much easier. Our experiments show that our method achieves higher segmentation accuracy with fewer semantic annotations than existing methods over a wide range of scenes.

Francesco Milano, Jen Jen Chung, Hermann Blum et al.

State-of-the-art approaches for 6D object pose estimation assume the availability of CAD models and require the user to manually set up physically-based rendering (PBR) pipelines for synthetic training data generation. Both factors limit the application of these methods in real-world scenarios. In this work, we present a pipeline that does not require CAD models and allows training a state-of-the-art pose estimator requiring only a small set of real images as input. Our method is based on a NeuS2 object representation, that we learn through a semi-automated procedure based on Structure-from-Motion (SfM) and object-agnostic segmentation. We exploit the novel-view synthesis ability of NeuS2 and simple cut-and-paste augmentation to automatically generate photorealistic object renderings, which we use to train the correspondence-based SurfEmb pose estimator. We evaluate our method on the LINEMOD-Occlusion dataset, extensively studying the impact of its individual components and showing competitive performance with respect to approaches based on CAD models and PBR data. We additionally demonstrate the ease of use and effectiveness of our pipeline on self-collected real-world objects, showing that our method outperforms state-of-the-art CAD-model-free approaches, with better accuracy and robustness to mild occlusions. To allow the robotics community to benefit from this system, we will publicly release it at https://www.github.com/ethz-asl/neusurfemb.

Michel Breyer, Jen Jen Chung, Lionel Ott et al.

General robot grasping in clutter requires the ability to synthesize grasps that work for previously unseen objects and that are also robust to physical interactions, such as collisions with other objects in the scene. In this work, we design and train a network that predicts 6 DOF grasps from 3D scene information gathered from an on-board sensor such as a wrist-mounted depth camera. Our proposed Volumetric Grasping Network (VGN) accepts a Truncated Signed Distance Function (TSDF) representation of the scene and directly outputs the predicted grasp quality and the associated gripper orientation and opening width for each voxel in the queried 3D volume. We show that our approach can plan grasps in only 10 ms and is able to clear 92% of the objects in real-world clutter removal experiments without the need for explicit collision checking. The real-time capability opens up the possibility for closed-loop grasp planning, allowing robots to handle disturbances, recover from errors and provide increased robustness. Code is available at https://github.com/ethz-asl/vgn.

Rik Bähnemann, Nicholas Lawrance, Jen Jen Chung et al.

In this paper, we present a path planner for low-altitude terrain coverage in known environments with unmanned rotary-wing micro aerial vehicles (MAVs). Airborne systems can assist humanitarian demining by surveying suspected hazardous areas (SHAs) with cameras, ground-penetrating synthetic aperture radar (GPSAR), and metal detectors. Most available coverage planner implementations for MAVs do not consider obstacles and thus cannot be deployed in obstructed environments. We describe an open source framework to perform coverage planning in polygon flight corridors with obstacles. Our planner extends boustrophedon coverage planning by optimizing over different sweep combinations to find the optimal sweep path, and considers obstacles during transition flights between cells. We evaluate the path planner on 320 synthetic maps and show that it is able to solve realistic planning instances fast enough to run in the field. The planner achieves 14% lower path costs than a conventional coverage planner. We validate the planner on a real platform where we show low-altitude coverage over a sloped terrain with trees.

Francesco Milano, Jen Jen Chung, Lionel Ott et al.

Surface normal integration is a fundamental problem in computer vision, dealing with the objective of reconstructing a surface from its corresponding normal map. Existing approaches require an iterative global optimization to jointly estimate the depth of each pixel, which scales poorly to larger normal maps. In this paper, we address this problem by recasting normal integration as the estimation of relative scales of continuous components. By constraining pixels belonging to the same component to jointly vary their scale, we drastically reduce the number of optimization variables. Our framework includes a heuristic to accurately estimate continuous components from the start, a strategy to rebalance optimization terms, and a technique to iteratively merge components to further reduce the size of the problem. Our method achieves state-of-the-art results on the standard normal integration benchmark in as little as a few seconds and achieves one-order-of-magnitude speedup over pixel-level approaches on large-resolution normal maps.

Florian Achermann, Thomas Stastny, Bogdan Danciu et al.

Real-time high-resolution wind predictions are beneficial for various applications including safe manned and unmanned aviation. Current weather models require too much compute and lack the necessary predictive capabilities as they are valid only at the scale of multiple kilometers and hours - much lower spatial and temporal resolutions than these applications require. Our work, for the first time, demonstrates the ability to predict low-altitude wind in real-time on limited-compute devices, from only sparse measurement data. We train a neural network, WindSeer, using only synthetic data from computational fluid dynamics simulations and show that it can successfully predict real wind fields over terrain with known topography from just a few noisy and spatially clustered wind measurements. WindSeer can generate accurate predictions at different resolutions and domain sizes on previously unseen topography without retraining. We demonstrate that the model successfully predicts historical wind data collected by weather stations and wind measured onboard drones.

Kenneth Blomqvist, Jen Jen Chung, Lionel Ott et al.

Creating computer vision datasets requires careful planning and lots of time and effort. In robotics research, we often have to use standardized objects, such as the YCB object set, for tasks such as object tracking, pose estimation, grasping and manipulation, as there are datasets and pre-learned methods available for these objects. This limits the impact of our research since learning-based computer vision methods can only be used in scenarios that are supported by existing datasets. In this work, we present a full object keypoint tracking toolkit, encompassing the entire process from data collection, labeling, model learning and evaluation. We present a semi-automatic way of collecting and labeling datasets using a wrist mounted camera on a standard robotic arm. Using our toolkit and method, we are able to obtain a working 3D object keypoint detector and go through the whole process of data collection, annotation and learning in just a couple hours of active time.

Rik Bähnemann, Nicholas Lawrance, Lucas Streichenberg et al.

Ground penetrating radar mounted on micro aerial vehicle (MAV) is a promising tool to assist humanitarian landmine clearance. However, the quality of synthetic aperture radar images depends on accurate and precise motion estimation of the radar antennas as well as generating informative viewpoints with the MAV. This paper presents a complete and automatic airborne ground-penetrating synthetic aperture radar (GPSAR) system. The system consists of a spatially calibrated and temporally synchronized industrial grade sensor suite that enables navigation above ground level, radar imaging, and optical imaging. A custom mission planning framework allows generation and automatic execution of stripmap and circular (GPSAR) trajectories controlled above ground level as well as aerial imaging survey flights. A factor graph based state estimator fuses measurements from dual receiver real-time kinematic (RTK) global navigation satellite system (GNSS) and inertial measurement unit (IMU) to obtain precise, high rate platform positions and orientations. Ground truth experiments showed sensor timing as accurate as 0.8 us and as precise as 0.1 us with localization rates of 1 kHz. The dual position factor formulation improves online localization accuracy up to 40% and batch localization accuracy up to 59% compared to a single position factor with uncertain heading initialization. Our field trials validated a localization accuracy and precision that enables coherent radar measurement addition and detection of radar targets buried in sand. This validates the potential as an aerial landmine detection system.

Jiapeng Zhong, Zheyu Ye, Andrei Cramariuc et al.

Many mobile robotic platforms rely on an accurate knowledge of the extrinsic calibration parameters, especially systems performing visual stereo matching. Although a number of accurate stereo camera calibration methods have been developed, which provide good initial "factory" calibrations, the determined parameters can lose their validity over time as the sensors are exposed to environmental conditions and external effects. Thus, on autonomous platforms on-board diagnostic methods for an early detection of the need to repeat calibration procedures have the potential to prevent critical failures of crucial systems, such as state estimation or obstacle detection. In this work, we present a novel data-driven method to estimate the calibration quality and detect discrepancies between the original calibration and the current system state for stereo camera systems. The framework consists of a novel dataset generation pipeline to train CalQNet, a deep convolutional neural network. CalQNet can estimate the calibration quality using a new metric that approximates the degree of miscalibration in stereo setups. We show the framework's ability to predict from a single stereo frame if a state-of-the-art stereo-visual odometry system will diverge due to a degraded calibration in two real-world experiments.

Hai Zhu, Jen Jen Chung, Nicholas R. J. Lawrance et al.

This paper presents an online informative path planning approach for active information gathering on three-dimensional surfaces using aerial robots. Most existing works on surface inspection focus on planning a path offline that can provide full coverage of the surface, which inherently assumes the surface information is uniformly distributed hence ignoring potential spatial correlations of the information field. In this paper, we utilize manifold Gaussian processes (mGPs) with geodesic kernel functions for mapping surface information fields and plan informative paths online in a receding horizon manner. Our approach actively plans information-gathering paths based on recent observations that respect dynamic constraints of the vehicle and a total flight time budget. We provide planning results for simulated temperature modeling for simple and complex 3D surface geometries (a cylinder and an aircraft model). We demonstrate that our informative planning method outperforms traditional approaches such as 3D coverage planning and random exploration, both in reconstruction error and information-theoretic metrics. We also show that by taking spatial correlations of the information field into planning using mGPs, the information gathering efficiency is significantly improved.

Daniel Dugas, Juan Nieto, Roland Siegwart et al.

Robot navigation is a task where reinforcement learning approaches are still unable to compete with traditional path planning. State-of-the-art methods differ in small ways, and do not all provide reproducible, openly available implementations. This makes comparing methods a challenge. Recent research has shown that unsupervised learning methods can scale impressively, and be leveraged to solve difficult problems. In this work, we design ways in which unsupervised learning can be used to assist reinforcement learning for robot navigation. We train two end-to-end, and 18 unsupervised-learning-based architectures, and compare them, along with existing approaches, in unseen test cases. We demonstrate our approach working on a real life robot. Our results show that unsupervised learning methods are competitive with end-to-end methods. We also highlight the importance of various components such as input representation, predictive unsupervised learning, and latent features. We make all our models publicly available, as well as training and testing environments, and tools. This release also includes OpenAI-gym-compatible environments designed to emulate the training conditions described by other papers, with as much fidelity as possible. Our hope is that this helps in bringing together the field of RL for robot navigation, and allows meaningful comparisons across state-of-the-art methods.

Le Chen, Yunke Ao, Florian Tschopp et al.

Visual-inertial systems rely on precise calibrations of both camera intrinsics and inter-sensor extrinsics, which typically require manually performing complex motions in front of a calibration target. In this work we present a novel approach to obtain favorable trajectories for visual-inertial system calibration, using model-based deep reinforcement learning. Our key contribution is to model the calibration process as a Markov decision process and then use model-based deep reinforcement learning with particle swarm optimization to establish a sequence of calibration trajectories to be performed by a robot arm. Our experiments show that while maintaining similar or shorter path lengths, the trajectories generated by our learned policy result in lower calibration errors compared to random or handcrafted trajectories.

Julian Förster, Lionel Ott, Juan Nieto et al.

Symbolic planning can provide an intuitive interface for non-expert users to operate autonomous robots by abstracting away much of the low-level programming. However, symbolic planners assume that the initially provided abstract domain and problem descriptions are closed and complete. This means that they are fundamentally unable to adapt to changes in the environment or task that are not captured by the initial description. We propose a method that allows an agent to automatically extend its skill set, and thus the abstract description, upon encountering such a situation. We introduce strategies for generalizing from previous experience, completing sequences of key actions and discovering preconditions to ensure the efficiency of our skill sequence exploration scheme. The resulting system is evaluated in simulation on object rearrangement tasks. Compared to a Monte Carlo Tree Search baseline, our strategies for efficient search have on average a 29% higher success rate at a 68% faster runtime.

Álvaro Serra-Gómez, Bruno Brito, Hai Zhu et al.

Decentralized multi-robot systems typically perform coordinated motion planning by constantly broadcasting their intentions as a means to cope with the lack of a central system coordinating the efforts of all robots. Especially in complex dynamic environments, the coordination boost allowed by communication is critical to avoid collisions between cooperating robots. However, the risk of collision between a pair of robots fluctuates through their motion and communication is not always needed. Additionally, constant communication makes much of the still valuable information shared in previous time steps redundant. This paper presents an efficient communication method that solves the problem of "when" and with "whom" to communicate in multi-robot collision avoidance scenarios. In this approach, every robot learns to reason about other robots' states and considers the risk of future collisions before asking for the trajectory plans of other robots. We evaluate and verify the proposed communication strategy in simulation with four quadrotors and compare it with three baseline strategies: non-communicating, broadcasting and a distance-based method broadcasting information with quadrotors within a predefined distance.

Kenneth Blomqvist, Michel Breyer, Andrei Cramariuc et al.

With humankind facing new and increasingly large-scale challenges in the medical and domestic spheres, automation of the service sector carries a tremendous potential for improved efficiency, quality, and safety of operations. Mobile robotics can offer solutions with a high degree of mobility and dexterity, however these complex systems require a multitude of heterogeneous components to be carefully integrated into one consistent framework. This work presents a mobile manipulation system that combines perception, localization, navigation, motion planning and grasping skills into one common workflow for fetch and carry applications in unstructured indoor environments. The tight integration across the various modules is experimentally demonstrated on the task of finding a commonly available object in an office environment, grasping it, and delivering it to a desired drop-off location. The accompanying video is available at https://youtu.be/e89_Xg1sLnY.

Leiv Andresen, Adrian Brandemuehl, Alex Hönger et al.

This paper presents the perception, mapping, and planning pipeline implemented on an autonomous race car. It was developed by the 2019 AMZ driverless team for the Formula Student Germany (FSG) 2019 driverless competition, where it won 1st place overall. The presented solution combines early fusion of camera and LiDAR data, a layered mapping approach, and a planning approach that uses Bayesian filtering to achieve high-speed driving on unknown race tracks while creating accurate maps. We benchmark the method against our team's previous solution, which won FSG 2018, and show improved accuracy when driving at the same speeds. Furthermore, the new pipeline makes it possible to reliably raise the maximum driving speed in unknown environments from 3~m/s to 12~m/s while still mapping with an acceptable RMSE of 0.29~m.

Julien Kindle, Fadri Furrer, Tonci Novkovic et al.

Mobile manipulation is usually achieved by sequentially executing base and manipulator movements. This simplification, however, leads to a loss in efficiency and in some cases a reduction of workspace size. Even though different methods have been proposed to solve Whole-Body Control (WBC) online, they are either limited by a kinematic model or do not allow for reactive, online obstacle avoidance. In order to overcome these drawbacks, in this work, we propose an end-to-end Reinforcement Learning (RL) approach to WBC. We compared our learned controller against a state-of-the-art sampling-based method in simulation and achieved faster overall mission times. In addition, we validated the learned policy on our mobile manipulator RoyalPanda in challenging narrow corridor environments.

Margarita Grinvald, Fadri Furrer, Tonci Novkovic et al.

To autonomously navigate and plan interactions in real-world environments, robots require the ability to robustly perceive and map complex, unstructured surrounding scenes. Besides building an internal representation of the observed scene geometry, the key insight toward a truly functional understanding of the environment is the usage of higher-level entities during mapping, such as individual object instances. We propose an approach to incrementally build volumetric object-centric maps during online scanning with a localized RGB-D camera. First, a per-frame segmentation scheme combines an unsupervised geometric approach with instance-aware semantic object predictions. This allows us to detect and segment elements both from the set of known classes and from other, previously unseen categories. Next, a data association step tracks the predicted instances across the different frames. Finally, a map integration strategy fuses information about their 3D shape, location, and, if available, semantic class into a global volume. Evaluation on a publicly available dataset shows that the proposed approach for building instance-level semantic maps is competitive with state-of-the-art methods, while additionally able to discover objects of unseen categories. The system is further evaluated within a real-world robotic mapping setup, for which qualitative results highlight the online nature of the method.

Marija Popovic, Teresa Vidal-Calleja, Jen Jen Chung et al.

Information gathering algorithms play a key role in unlocking the potential of robots for efficient data collection in a wide range of applications. However, most existing strategies neglect the fundamental problem of the robot pose uncertainty, which is an implicit requirement for creating robust, high-quality maps. To address this issue, we introduce an informative planning framework for active mapping that explicitly accounts for the pose uncertainty in both the mapping and planning tasks. Our strategy exploits a Gaussian Process (GP) model to capture a target environmental field given the uncertainty on its inputs. For planning, we formulate a new utility function that couples the localization and field mapping objectives in GP-based mapping scenarios in a principled way, without relying on any manually tuned parameters. Extensive simulations show that our approach outperforms existing strategies, with reductions in mean pose uncertainty and map error. We also present a proof of concept in an indoor temperature mapping scenario.

Marija Popovic, Teresa Vidal-Calleja, Gregory Hitz et al.

Unmanned Aerial Vehicles (UAVs) represent a new frontier in a wide range of monitoring and research applications. To fully leverage their potential, a key challenge is planning missions for efficient data acquisition in complex environments. To address this issue, this article introduces a general Informative Path Planning (IPP) framework for monitoring scenarios using an aerial robot, focusing on problems in which the value of sensor information is unevenly distributed in a target area and unknown a priori . The approach is capable of learning and focusing on regions of interest via adaptation to map either discrete or continuous variables on the terrain using variable-resolution data received from probabilistic sensors. During a mission, the terrain maps built online are used to plan information-rich trajectories in continuous 3-D space by optimizing initial solutions obtained by a coarse grid search. Extensive simulations show that our approach is more efficient than existing methods. We also demonstrate its real-time application on a photorealistic mapping scenario using a publicly available dataset and demonstrate a proof of concept for an agricultural monitoring task.