Błażej Osiński

Dhruv Shah, Blazej Osinski, Brian Ichter et al. · berkeley

Goal-conditioned policies for robotic navigation can be trained on large, unannotated datasets, providing for good generalization to real-world settings. However, particularly in vision-based settings where specifying goals requires an image, this makes for an unnatural interface. Language provides a more convenient modality for communication with robots, but contemporary methods typically require expensive supervision, in the form of trajectories annotated with language descriptions. We present a system, LM-Nav, for robotic navigation that enjoys the benefits of training on unannotated large datasets of trajectories, while still providing a high-level interface to the user. Instead of utilizing a labeled instruction following dataset, we show that such a system can be constructed entirely out of pre-trained models for navigation (ViNG), image-language association (CLIP), and language modeling (GPT-3), without requiring any fine-tuning or language-annotated robot data. We instantiate LM-Nav on a real-world mobile robot and demonstrate long-horizon navigation through complex, outdoor environments from natural language instructions. For videos of our experiments, code release, and an interactive Colab notebook that runs in your browser, please check out our project page https://sites.google.com/view/lmnav

Dhruv Shah, Michael Equi, Blazej Osinski et al. · berkeley

Navigation in unfamiliar environments presents a major challenge for robots: while mapping and planning techniques can be used to build up a representation of the world, quickly discovering a path to a desired goal in unfamiliar settings with such methods often requires lengthy mapping and exploration. Humans can rapidly navigate new environments, particularly indoor environments that are laid out logically, by leveraging semantics -- e.g., a kitchen often adjoins a living room, an exit sign indicates the way out, and so forth. Language models can provide robots with such knowledge, but directly using language models to instruct a robot how to reach some destination can also be impractical: while language models might produce a narrative about how to reach some goal, because they are not grounded in real-world observations, this narrative might be arbitrarily wrong. Therefore, in this paper we study how the ``semantic guesswork'' produced by language models can be utilized as a guiding heuristic for planning algorithms. Our method, Language Frontier Guide (LFG), uses the language model to bias exploration of novel real-world environments by incorporating the semantic knowledge stored in language models as a search heuristic for planning with either topological or metric maps. We evaluate LFG in challenging real-world environments and simulated benchmarks, outperforming uninformed exploration and other ways of using language models.

Yan Wu, Esther Wershof, Sebastian M Schmon et al.

We introduce a comprehensive framework for modeling single cell transcriptomic responses to perturbations, aimed at standardizing benchmarking in this rapidly evolving field. Our approach includes a modular and user-friendly model development and evaluation platform, a collection of diverse perturbational datasets, and a set of metrics designed to fairly compare models and dissect their performance. Through extensive evaluation of both published and baseline models across diverse datasets, we highlight the limitations of widely used models, such as mode collapse. We also demonstrate the importance of rank metrics which complement traditional model fit measures, such as RMSE, for validating model effectiveness. Notably, our results show that while no single model architecture clearly outperforms others, simpler architectures are generally competitive and scale well with larger datasets. Overall, this benchmarking exercise sets new standards for model evaluation, supports robust model development, and furthers the use of these models to simulate genetic and chemical screens for therapeutic discovery.

Zichao Yan, Yan Wu, Mica Xu Ji et al.

Predicting the effects of perturbations in-silico on cell state can identify drivers of cell behavior at scale and accelerate drug discovery. However, modeling challenges remain due to the inherent heterogeneity of single cell gene expression and the complex, latent gene dependencies. Here, we present PRiMeFlow, an end-to-end flow matching based approach to directly model the effects of genetic and small molecule perturbations in the gene expression space. The distribution-fitting approach taken by PRiMeFlow enables it to accurately approximate the empirical distribution of single-cell gene expression, which we demonstrate through extensive benchmarking inside PerturBench. Through ablation studies, we also validate important model design choices such as operating in gene expression space and parameterizing the velocity field with a U-Net architecture. The PRiMeFlow architecture was used as the basis for the model that won the Generalist Prize in the first ARC Virtual Cell Challenge.

Błażej Osiński, Piotr Miłoś, Adam Jakubowski et al.

This work introduces interactive traffic scenarios in the CARLA simulator, which are based on real-world traffic. We concentrate on tactical tasks lasting several seconds, which are especially challenging for current control methods. The CARLA Real Traffic Scenarios (CRTS) is intended to be a training and testing ground for autonomous driving systems. To this end, we open-source the code under a permissive license and present a set of baseline policies. CRTS combines the realism of traffic scenarios and the flexibility of simulation. We use it to train agents using a reinforcement learning algorithm. We show how to obtain competitive polices and evaluate experimentally how observation types and reward schemes affect the training process and the resulting agent's behavior.

Michał Zawalski, Błażej Osiński, Henryk Michalewski et al.

Multi-agent reinforcement learning (MARL) provides a framework for problems involving multiple interacting agents. Despite apparent similarity to the single-agent case, multi-agent problems are often harder to train and analyze theoretically. In this work, we propose MA-Trace, a new on-policy actor-critic algorithm, which extends V-Trace to the MARL setting. The key advantage of our algorithm is its high scalability in a multi-worker setting. To this end, MA-Trace utilizes importance sampling as an off-policy correction method, which allows distributing the computations with no impact on the quality of training. Furthermore, our algorithm is theoretically grounded - we prove a fixed-point theorem that guarantees convergence. We evaluate the algorithm extensively on the StarCraft Multi-Agent Challenge, a standard benchmark for multi-agent algorithms. MA-Trace achieves high performance on all its tasks and exceeds state-of-the-art results on some of them.

Matt Vitelli, Yan Chang, Yawei Ye et al.

In this paper we present the first safe system for full control of self-driving vehicles trained from human demonstrations and deployed in challenging, real-world, urban environments. Current industry-standard solutions use rule-based systems for planning. Although they perform reasonably well in common scenarios, the engineering complexity renders this approach incompatible with human-level performance. On the other hand, the performance of machine-learned (ML) planning solutions can be improved by simply adding more exemplar data. However, ML methods cannot offer safety guarantees and sometimes behave unpredictably. To combat this, our approach uses a simple yet effective rule-based fallback layer that performs sanity checks on an ML planner's decisions (e.g. avoiding collision, assuring physical feasibility). This allows us to leverage ML to handle complex situations while still assuring the safety, reducing ML planner-only collisions by 95%. We train our ML planner on 300 hours of expert driving demonstrations using imitation learning and deploy it along with the fallback layer in downtown San Francisco, where it takes complete control of a real vehicle and navigates a wide variety of challenging urban driving scenarios.

Oliver Scheel, Luca Bergamini, Maciej Wołczyk et al.

In this work we are the first to present an offline policy gradient method for learning imitative policies for complex urban driving from a large corpus of real-world demonstrations. This is achieved by building a differentiable data-driven simulator on top of perception outputs and high-fidelity HD maps of the area. It allows us to synthesize new driving experiences from existing demonstrations using mid-level representations. Using this simulator we then train a policy network in closed-loop employing policy gradients. We train our proposed method on 100 hours of expert demonstrations on urban roads and show that it learns complex driving policies that generalize well and can perform a variety of driving maneuvers. We demonstrate this in simulation as well as deploy our model to self-driving vehicles in the real-world. Our method outperforms previously demonstrated state-of-the-art for urban driving scenarios -- all this without the need for complex state perturbations or collecting additional on-policy data during training. We make code and data publicly available.

Long Chen, Lukas Platinsky, Stefanie Speichert et al.

We investigate what grade of sensor data is required for training an imitation-learning-based AV planner on human expert demonstration. Machine-learned planners are very hungry for training data, which is usually collected using vehicles equipped with the same sensors used for autonomous operation. This is costly and non-scalable. If cheaper sensors could be used for collection instead, data availability would go up, which is crucial in a field where data volume requirements are large and availability is small. We present experiments using up to 1000 hours worth of expert demonstration and find that training with 10x lower-quality data outperforms 1x AV-grade data in terms of planner performance. The important implication of this is that cheaper sensors can indeed be used. This serves to improve data access and democratize the field of imitation-based motion planning. Alongside this, we perform a sensitivity analysis of planner performance as a function of perception range, field-of-view, accuracy, and data volume, and the reason why lower-quality data still provide good planning results.

Luca Bergamini, Yawei Ye, Oliver Scheel et al.

In this work, we present a simple end-to-end trainable machine learning system capable of realistically simulating driving experiences. This can be used for the verification of self-driving system performance without relying on expensive and time-consuming road testing. In particular, we frame the simulation problem as a Markov Process, leveraging deep neural networks to model both state distribution and transition function. These are trainable directly from the existing raw observations without the need for any handcrafting in the form of plant or kinematic models. All that is needed is a dataset of historical traffic episodes. Our formulation allows the system to construct never seen scenes that unfold realistically reacting to the self-driving car's behaviour. We train our system directly from 1,000 hours of driving logs and measure both realism, reactivity of the simulation as the two key properties of the simulation. At the same time, we apply the method to evaluate the performance of a recently proposed state-of-the-art ML planning system trained from human driving logs. We discover this planning system is prone to previously unreported causal confusion issues that are difficult to test by non-reactive simulation. To the best of our knowledge, this is the first work that directly merges highly realistic data-driven simulations with a closed-loop evaluation for self-driving vehicles. We make the data, code, and pre-trained models publicly available to further stimulate simulation development.

Błażej Osiński, Adam Jakubowski, Piotr Miłoś et al.

We use reinforcement learning in simulation to obtain a driving system controlling a full-size real-world vehicle. The driving policy takes RGB images from a single camera and their semantic segmentation as input. We use mostly synthetic data, with labelled real-world data appearing only in the training of the segmentation network. Using reinforcement learning in simulation and synthetic data is motivated by lowering costs and engineering effort. In real-world experiments we confirm that we achieved successful sim-to-real policy transfer. Based on the extensive evaluation, we analyze how design decisions about perception, control, and training impact the real-world performance.

Lukasz Kaiser, Mohammad Babaeizadeh, Piotr Milos et al.

Model-free reinforcement learning (RL) can be used to learn effective policies for complex tasks, such as Atari games, even from image observations. However, this typically requires very large amounts of interaction -- substantially more, in fact, than a human would need to learn the same games. How can people learn so quickly? Part of the answer may be that people can learn how the game works and predict which actions will lead to desirable outcomes. In this paper, we explore how video prediction models can similarly enable agents to solve Atari games with fewer interactions than model-free methods. We describe Simulated Policy Learning (SimPLe), a complete model-based deep RL algorithm based on video prediction models and present a comparison of several model architectures, including a novel architecture that yields the best results in our setting. Our experiments evaluate SimPLe on a range of Atari games in low data regime of 100k interactions between the agent and the environment, which corresponds to two hours of real-time play. In most games SimPLe outperforms state-of-the-art model-free algorithms, in some games by over an order of magnitude.

Łukasz Kidziński, Sharada Prasanna Mohanty, Carmichael Ong et al.

In the NIPS 2017 Learning to Run challenge, participants were tasked with building a controller for a musculoskeletal model to make it run as fast as possible through an obstacle course. Top participants were invited to describe their algorithms. In this work, we present eight solutions that used deep reinforcement learning approaches, based on algorithms such as Deep Deterministic Policy Gradient, Proximal Policy Optimization, and Trust Region Policy Optimization. Many solutions use similar relaxations and heuristics, such as reward shaping, frame skipping, discretization of the action space, symmetry, and policy blending. However, each of the eight teams implemented different modifications of the known algorithms.