Chenran Li

Mayank Mittal, Pascal Roth, James Tigue et al. · nvidia

We present Isaac Lab, the natural successor to Isaac Gym, which extends the paradigm of GPU-native robotics simulation into the era of large-scale multi-modal learning. Isaac Lab combines high-fidelity GPU parallel physics, photorealistic rendering, and a modular, composable architecture for designing environments and training robot policies. Beyond physics and rendering, the framework integrates actuator models, multi-frequency sensor simulation, data collection pipelines, and domain randomization tools, unifying best practices for reinforcement and imitation learning at scale within a single extensible platform. We highlight its application to a diverse set of challenges, including whole-body control, cross-embodiment mobility, contact-rich and dexterous manipulation, and the integration of human demonstrations for skill acquisition. Finally, we discuss upcoming integration with the differentiable, GPU-accelerated Newton physics engine, which promises new opportunities for scalable, data-efficient, and gradient-based approaches to robot learning. We believe Isaac Lab's combination of advanced simulation capabilities, rich sensing, and data-center scale execution will help unlock the next generation of breakthroughs in robotics research.

Yiheng Li, Seth Z. Zhao, Chenfeng Xu et al.

Accumulating substantial volumes of real-world driving data proves pivotal in the realm of trajectory forecasting for autonomous driving. Given the heavy reliance of current trajectory forecasting models on data-driven methodologies, we aim to tackle the challenge of learning general trajectory forecasting representations under limited data availability. We propose a pipeline-level solution to mitigate the issue of data scarcity in trajectory forecasting. The solution is composed of two parts: firstly, we adopt HD map augmentation and trajectory synthesis for generating driving data, and then we learn representations by pre-training on them. Specifically, we apply vector transformations to reshape the maps, and then employ a rule-based model to generate trajectories on both original and augmented scenes; thus enlarging the driving data without collecting additional real ones. To foster the learning of general representations within this augmented dataset, we comprehensively explore the different pre-training strategies, including extending the concept of a Masked AutoEncoder (MAE) for trajectory forecasting. Without bells and whistles, our proposed pipeline-level solution is general, simple, yet effective: we conduct extensive experiments to demonstrate the effectiveness of our data expansion and pre-training strategies, which outperform the baseline prediction model by large margins, e.g. 5.04%, 3.84% and 8.30% in terms of $MR_6$, $minADE_6$ and $minFDE_6$. The pre-training dataset and the codes for pre-training and fine-tuning are released at https://github.com/yhli123/Pretraining_on_Synthetic_Driving_Data_for_Trajectory_Prediction.

Wei-Jer Chang, Chen Tang, Chenran Li et al.

Traffic simulation plays a crucial role in evaluating and improving autonomous driving planning systems. After being deployed on public roads, autonomous vehicles need to interact with human road participants with different social preferences (e.g., selfish or courteous human drivers). To ensure that autonomous vehicles take safe and efficient maneuvers in different interactive traffic scenarios, we should be able to evaluate autonomous vehicles against reactive agents with different social characteristics in the simulation environment. We propose a socially-controllable behavior generation (SCBG) model for this purpose, which allows the users to specify the level of courtesy of the generated trajectory while ensuring realistic and human-like trajectory generation through learning from real-world driving data. Specifically, we define a novel and differentiable measure to quantify the level of courtesy of driving behavior, leveraging marginal and conditional behavior prediction models trained from real-world driving data. The proposed courtesy measure allows us to auto-label the courtesy levels of trajectories from real-world driving data and conveniently train an SCBG model generating trajectories based on the input courtesy values. We examined the SCBG model on the Waymo Open Motion Dataset (WOMD) and showed that we were able to control the SCBG model to generate realistic driving behaviors with desired courtesy levels. Interestingly, we found that the SCBG model was able to identify different motion patterns of courteous behaviors according to the scenarios.

Chenran Li, Chen Tang, Haruki Nishimura et al.

Imitation Learning (IL) is a widely used framework for learning imitative behavior from demonstrations. It is especially appealing for solving complex real-world tasks where handcrafting reward function is difficult, or when the goal is to mimic human expert behavior. However, the learned imitative policy can only follow the behavior in the demonstration. When applying the imitative policy, we may need to customize the policy behavior to meet different requirements coming from diverse downstream tasks. Meanwhile, we still want the customized policy to maintain its imitative nature. To this end, we formulate a new problem setting called policy customization. It defines the learning task as training a policy that inherits the characteristics of the prior policy while satisfying some additional requirements imposed by a target downstream task. We propose a novel and principled approach to interpret and determine the trade-off between the two task objectives. Specifically, we formulate the customization problem as a Markov Decision Process (MDP) with a reward function that combines 1) the inherent reward of the demonstration; and 2) the add-on reward specified by the downstream task. We propose a novel framework, Residual Q-learning, which can solve the formulated MDP by leveraging the prior policy without knowing the inherent reward or value function of the prior policy. We derive a family of residual Q-learning algorithms that can realize offline and online policy customization, and show that the proposed algorithms can effectively accomplish policy customization tasks in various environments. Demo videos and code are available on our website: https://sites.google.com/view/residualq-learning.

Tingwu Wang, Olivier Dionne, Michael De Ruyter et al.

Despite transformative advances in generative motion synthesis, real-time interactive motion control remains dominated by traditional techniques. In this work, we identify two key challenges in bridging research and production: 1) Real-time scalability: Industry applications demand real-time generation of a vast repertoire of motion skills, while generative methods exhibit significant degradation in quality and scalability under real-time computation constraints, and 2) Integration: Industry applications demand fine-grained multi-modal control involving velocity commands, style selection, and precise keyframes, a need largely unmet by existing text- or tag-driven models. To overcome these limitations, we introduce MotionBricks: a large-scale, real-time generative framework with a two-fold solution. First, we propose a large-scale modular latent generative backbone tailored for robust real-time motion generation, effectively modeling a dataset of over 350,000 motion clips with a single model. Second, we introduce smart primitives that provide a unified, robust, and intuitive interface for authoring both navigation and object interaction. Applications can be designed in a plug-and-play manner like assembling bricks without expert animation knowledge. Quantitatively, we show that MotionBricks produces state-of-the-art motion quality on open-source and proprietary datasets of various scales, while also achieving a real-time throughput of 15,000 FPS with 2ms latency. We demonstrate the flexibility and robustness of MotionBricks in a complete production-level animation demo, covering navigation and object-scene interaction across various styles with a unified model. To showcase our framework's application beyond animation, we deploy MotionBricks on the Unitree G1 humanoid robot to demonstrate its flexibility and generalization for real-time robotic control.

Wei-Jer Chang, Yeping Hu, Chenran Li et al.

Simulation has played an important role in efficiently evaluating self-driving vehicles in terms of scalability. Existing methods mostly rely on heuristic-based simulation, where traffic participants follow certain human-encoded rules that fail to generate complex human behaviors. Therefore, the reactive simulation concept is proposed to bridge the human behavior gap between simulation and real-world traffic scenarios by leveraging real-world data. However, these reactive models can easily generate unreasonable behaviors after a few steps of simulation, where we regard the model as losing its stability. To the best of our knowledge, no work has explicitly discussed and analyzed the stability of the reactive simulation framework. In this paper, we aim to provide a thorough stability analysis of the reactive simulation and propose a solution to enhance the stability. Specifically, we first propose a new reactive simulation framework, where we discover that the smoothness and consistency of the simulated state sequences are crucial factors to stability. We then incorporate the kinematic vehicle model into the framework to improve the closed-loop stability of the reactive simulation. Furthermore, along with commonly-used metrics, several novel metrics are proposed in this paper to better analyze the simulation performance.

Yuxin Chen, Chen Tang, Ran Tian et al.

Generalization poses a significant challenge in Multi-agent Reinforcement Learning (MARL). The extent to which an agent is influenced by unseen co-players depends on the agent's policy and the specific scenario. A quantitative examination of this relationship sheds light on effectively training agents for diverse scenarios. In this study, we present the Level of Influence (LoI), a metric quantifying the interaction intensity among agents within a given scenario and environment. We observe that, generally, a more diverse set of co-play agents during training enhances the generalization performance of the ego agent; however, this improvement varies across distinct scenarios and environments. LoI proves effective in predicting these improvement disparities within specific scenarios. Furthermore, we introduce a LoI-guided resource allocation method tailored to train a set of policies for diverse scenarios under a constrained budget. Our results demonstrate that strategic resource allocation based on LoI can achieve higher performance than uniform allocation under the same computation budget.

Zhengyi Luo, Ye Yuan, Tingwu Wang et al.

Despite the rise of billion-parameter foundation models trained across thousands of GPUs, similar scaling gains have not been shown for humanoid control. Current neural controllers for humanoids remain modest in size, target a limited behavior set, and are trained on a handful of GPUs over several days. We show that scaling up model capacity, data, and compute yields a generalist humanoid controller capable of creating natural and robust whole-body movements. Specifically, we posit motion tracking as a natural and scalable task for humanoid control, leverageing dense supervision from diverse motion-capture data to acquire human motion priors without manual reward engineering. We build a foundation model for motion tracking by scaling along three axes: network size (from 1.2M to 42M parameters), dataset volume (over 100M frames, 700 hours of high-quality motion data), and compute (9k GPU hours). Beyond demonstrating the benefits of scale, we show the practical utility of our model through two mechanisms: (1) a real-time universal kinematic planner that bridges motion tracking to downstream task execution, enabling natural and interactive control, and (2) a unified token space that supports various motion input interfaces, such as VR teleoperation devices, human videos, and vision-language-action (VLA) models, all using the same policy. Scaling motion tracking exhibits favorable properties: performance improves steadily with increased compute and data diversity, and learned representations generalize to unseen motions, establishing motion tracking at scale as a practical foundation for humanoid control.

Chenran Li, Aolin Xu, Enna Sachdeva et al.

The driver warning system that alerts the human driver about potential risks during driving is a key feature of an advanced driver assistance system. Existing driver warning technologies, mainly the forward collision warning and unsafe lane change warning, can reduce the risk of collision caused by human errors. However, the current design methods have several major limitations. Firstly, the warnings are mainly generated in a one-shot manner without modeling the ego driver's reactions and surrounding objects, which reduces the flexibility and generality of the system over different scenarios. Additionally, the triggering conditions of warning are mostly rule-based threshold-checking given the current state, which lacks the prediction of the potential risk in a sufficiently long future horizon. In this work, we study the problem of optimally generating driver warnings by considering the interactions among the generated warning, the driver behavior, and the states of ego and surrounding vehicles on a long horizon. The warning generation problem is formulated as a partially observed Markov decision process (POMDP). An optimal warning generation framework is proposed as a solution to the proposed POMDP. The simulation experiments demonstrate the superiority of the proposed solution to the existing warning generation methods.

Pengcheng Wang, Xinghao Zhu, Yuxin Chen et al.

Reinforcement Learning and Imitation Learning have achieved widespread success in many domains but remain constrained during real-world deployment. One of the main issues is the additional requirements that were not considered during training. To address this challenge, policy customization has been introduced, aiming to adapt a prior policy while preserving its inherent properties and meeting new task-specific requirements. A principled approach to policy customization is Residual Q-Learning (RQL), which formulates the problem as a Markov Decision Process (MDP) and derives a family of value-based learning algorithms. However, RQL has not yet been applied to policy gradient methods, which restricts its applicability, especially in tasks where policy gradient has already proven more effective. In this work, we first derive a concise form of Soft Policy Gradient as a preliminary. Building on this, we introduce Residual Policy Gradient (RPG), which extends RQL to policy gradient methods, allowing policy customization in gradient-based RL settings. With the view of RPG, we rethink the KL-regularized objective widely used in RL fine-tuning. We show that under certain assumptions, KL-regularized objective leads to a maximum-entropy policy that balances the inherent properties and task-specific requirements on a reward-level. Our experiments in MuJoCo demonstrate the effectiveness of Soft Policy Gradient and Residual Policy Gradient.

Sirui Chen, Zi-ang Cao, Zhengyi Luo et al.

Recent progress in humanoid robots has unlocked agile locomotion skills, including backflipping, running, and crawling. Yet it remains challenging for a humanoid robot to perform forceful manipulation tasks such as moving objects, wiping, and pushing a cart. We propose adaptive Compliance Humanoid control through hIsight Perturbation (CHIP), a plug-and-play module that enables controllable end-effector stiffness while preserving agile tracking of dynamic reference motions. CHIP is easy to implement and requires neither data augmentation nor additional reward tuning. We show that a generalist motion-tracking controller trained with CHIP can perform a diverse set of forceful manipulation tasks that require different end-effector compliance, such as multi-robot collaboration, wiping, box delivery, and door opening.

Yuxin Chen, Chen Tang, Jianglan Wei et al.

Aligning robot behavior with human preferences is crucial for deploying embodied AI agents in human-centered environments. A promising solution is interactive imitation learning from human intervention, where a human expert observes the policy's execution and provides interventions as feedback. However, existing methods often fail to utilize the prior policy efficiently to facilitate learning, thus hindering sample efficiency. In this work, we introduce MEReQ (Maximum-Entropy Residual-Q Inverse Reinforcement Learning), designed for sample-efficient alignment from human intervention. Instead of inferring the complete human behavior characteristics, MEReQ infers a residual reward function that captures the discrepancy between the human expert's and the prior policy's underlying reward functions. It then employs Residual Q-Learning (RQL) to align the policy with human preferences using this residual reward function. Extensive evaluations on simulated and real-world tasks demonstrate that MEReQ achieves sample-efficient policy alignment from human intervention.

Zhao-Heng Yin, Chenran Li, Liting Sun et al.

We propose an imitation learning system for autonomous driving in urban traffic with interactions. We train a Behavioral Cloning~(BC) policy to imitate driving behavior collected from the real urban traffic, and apply the data aggregation algorithm to improve its performance iteratively. Applying data aggregation in this setting comes with two challenges. The first challenge is that it is expensive and dangerous to collect online rollout data in the real urban traffic. Creating similar traffic scenarios in simulator like CARLA for online rollout collection can also be difficult. Instead, we propose to create a weak simulator from the training dataset, in which all the surrounding vehicles follow the data trajectory provided by the dataset. We find that the collected online data in such a simulator can still be used to improve BC policy's performance. The second challenge is the tedious and time-consuming process of human labelling process during online rollout. To solve this problem, we use an A$^*$ planner as a pseudo-expert to provide expert-like demonstration. We validate our proposed imitation learning system in the real urban traffic scenarios. The experimental results show that our system can significantly improve the performance of baseline BC policy.