Harish Ravichandar

Abhineet Jain, Jack Kolb, Harish Ravichandar

Existing learning approaches to dexterous manipulation use demonstrations or interactions with the environment to train black-box neural networks that provide little control over how the robot learns the skills or how it would perform post training. These approaches pose significant challenges when implemented on physical platforms given that, during initial stages of training, the robot's behavior could be erratic and potentially harmful to its own hardware, the environment, or any humans in the vicinity. A potential way to address these limitations is to add constraints during learning that restrict and guide the robot's behavior during training as well as roll outs. Inspired by the success of constrained approaches in other domains, we investigate the effects of adding position-based constraints to a 24-DOF robot hand learning to perform object relocation using Constrained Policy Optimization. We find that a simple geometric constraint can ensure the robot learns to move towards the object sooner than without constraints. Further, training with this constraint requires a similar number of samples as its unconstrained counterpart to master the skill. These findings shed light on how simple constraints can help robots achieve sensible and safe behavior quickly and ease concerns surrounding hardware deployment. We also investigate the effects of the strictness of these constraints and report findings that provide insights into how different degrees of strictness affect learning outcomes. Our code is available at https://github.com/GT-STAR-Lab/constrained-rl-dexterous-manipulation.

Zhaodong Yang, Ai-Ping Hu, Harish Ravichandar

Automating chicken shoulder deboning requires precise 6-DoF cutting through a partially occluded, deformable, multi-material joint, since contact with the bones presents serious health and safety risks. Our work makes both systems-level and algorithmic contributions to train and deploy a reactive force-feedback cutting policy that dynamically adapts a nominal trajectory and enables full 6-DoF knife control to traverse the narrow joint gap while avoiding contact with the bones. First, we introduce an open-source custom-built simulator for multi-material cutting that models coupling, fracture, and cutting forces, and supports reinforcement learning, enabling efficient training and rapid prototyping. Second, we design a reusable physical testbed to emulate the chicken shoulder: two rigid "bone" spheres with controllable pose embedded in a softer block, enabling rigorous and repeatable evaluation while preserving essential multi-material characteristics of the target problem. Third, we train and deploy a residual RL policy, with discretized force observations and domain randomization, enabling robust zero-shot sim-to-real transfer and the first demonstration of a learned policy that debones a real chicken shoulder. Our experiments in our simulator, on our physical testbed, and on real chicken shoulders show that our learned policy reliably navigates the joint gap and reduces undesired bone/cartilage contact, resulting in up to a 4x improvement over existing open-loop cutting baselines in terms of success rate and bone avoidance. Our results also illustrate the necessity of force feedback for safe and effective multi-material cutting. The project website is at https://hal-zhaodong-yang.github.io/MultiMaterialWebsite/.

Abhineet Jain, Jack Kolb, J. M. Abbess et al.

Imitation learning is a promising approach to help robots acquire dexterous manipulation capabilities without the need for a carefully-designed reward or a significant computational effort. However, existing imitation learning approaches require sophisticated data collection infrastructure and struggle to generalize beyond the training distribution. One way to address this limitation is to gather additional data that better represents the full operating conditions. In this work, we investigate characteristics of such additional demonstrations and their impact on performance. Specifically, we study the effects of corrective and randomly-sampled additional demonstrations on learning a policy that guides a five-fingered robot hand through a pick-and-place task. Our results suggest that corrective demonstrations considerably outperform randomly-sampled demonstrations, when the proportion of additional demonstrations sampled from the full task distribution is larger than the number of original demonstrations sampled from a restrictive training distribution. Conversely, when the number of original demonstrations are higher than that of additional demonstrations, we find no significant differences between corrective and randomly-sampled additional demonstrations. These results provide insights into the inherent trade-off between the effort required to collect corrective demonstrations and their relative benefits over randomly-sampled demonstrations. Additionally, we show that inexpensive vision-based sensors, such as LeapMotion, can be used to dramatically reduce the cost of providing demonstrations for dexterous manipulation tasks. Our code is available at https://github.com/GT-STAR-Lab/corrective-demos-dexterous-manipulation.

Hrishikesh Sathyanarayan, Victor Vantilborgh, Harish Ravichandar et al.

Reliable robotic manipulation requires control policies that can accurately represent and adapt to uncertainty arising from contact-rich interactions. Modern data-driven methods mitigate uncertainty through large-scale training and computation, and degrade significantly in performance with limited number of training samples. By contrast, classical model-based controllers are computationally efficient and reliable, but their limited ability to represent task-relevant uncertainty can hinder performance in contact-rich interactions. In this work, we propose to expand the capabilities of model-based manipulation control through more flexible uncertainty modeling that retains performance while exactly adapting to uncertainty. Our approach casts the manipulation problem as a distributionally robust control optimization and proposes a novel deterministic formulation based on Stein variational inference that preserves performance while explicitly modeling task-sensitive parameter uncertainty. As a result, the derived controllers are more aware of task sensitivities to uncertainty, yielding high reliability without compromising performance. Experimental results demonstrate up to 3$\times$ improved robustness across a range of contact-rich manipulation tasks under broad parametric uncertainty, outperforming existing model-based control methods.

Shalin Anand Jain, Jiazhen Liu, Siva Kailas et al.

Multi-agent reinforcement learning (MARL) has emerged as a promising solution for learning complex and scalable coordination behaviors in multi-robot systems. However, established MARL platforms (e.g., SMAC and MPE) lack robotics relevance and hardware deployment, leaving multi-robot learning researchers to develop bespoke environments and hardware testbeds dedicated to the development and evaluation of their individual contributions. The Multi-Agent RL Benchmark and Learning Environment for the Robotarium (MARBLER) is an exciting recent step in providing a standardized robotics-relevant platform for MARL, by bridging the Robotarium testbed with existing MARL software infrastructure. However, MARBLER lacks support for parallelization and GPU/TPU execution, making the platform prohibitively slow compared to modern MARL environments and hindering adoption. We contribute JaxRobotarium, a Jax-powered end-to-end simulation, learning, deployment, and benchmarking platform for the Robotarium. JaxRobotarium enables rapid training and deployment of multi-robot RL (MRRL) policies with realistic robot dynamics and safety constraints, supporting parallelization and hardware acceleration. Our generalizable learning interface integrates easily with SOTA MARL libraries (e.g., JaxMARL). In addition, JaxRobotarium includes eight standardized coordination scenarios, including four novel scenarios that bring established MARL benchmark tasks (e.g., RWARE and Level-Based Foraging) to a robotics setting. We demonstrate that JaxRobotarium retains high simulation fidelity while achieving dramatic speedups over baseline (20x in training and 150x in simulation), and provides an open-access sim-to-real evaluation pipeline through the Robotarium testbed, accelerating and democratizing access to multi-robot learning research and evaluation. Our code is available at https://github.com/GT-STAR-Lab/JaxRobotarium.

Siva Kailas, Shalin Jain, Harish Ravichandar

Graph Neural Networks (GNNs), developed by the graph learning community, have been adopted and shown to be highly effective in multi-robot and multi-agent learning. Inspired by this successful cross-pollination, we investigate and characterize the suitability of existing GNN explanation methods for explaining multi-agent coordination. We find that these methods have the potential to identify the most-influential communication channels that impact the team's behavior. Informed by our initial analyses, we propose an attention entropy regularization term that renders GAT-based policies more amenable to existing graph-based explainers. Intuitively, minimizing attention entropy incentivizes agents to limit their attention to the most influential or impactful agents, thereby easing the challenge faced by the explainer. We theoretically ground this intuition by showing that minimizing attention entropy increases the disparity between the explainer-generated subgraph and its complement. Evaluations across three tasks and three team sizes i) provides insights into the effectiveness of existing explainers, and ii) demonstrates that our proposed regularization consistently improves explanation quality without sacrificing task performance.

Kevin Fu, Shalin Anand Jain, Pierce Howell et al.

Recent advances have enabled heterogeneous multi-robot teams to learn complex and effective coordination skills. However, existing neural architectures that support heterogeneous teaming tend to force a trade-off between expressivity and efficiency. Shared-parameter designs prioritize sample efficiency by enabling a single network to be shared across all or a pre-specified subset of robots (via input augmentations), but tend to limit behavioral diversity. In contrast, recent designs employ a separate policy for each robot, enabling greater diversity and expressivity at the cost of efficiency and generalization. Our key insight is that such tradeoffs can be avoided by viewing these design choices as ends of a broad spectrum. Inspired by recent work in transfer and meta learning, and building on prior work in multi-robot task allocation, we propose Capability-Aware Shared Hypernetworks (CASH), a soft weight sharing architecture that uses hypernetworks to efficiently learn a flexible shared policy that dynamically adapts to each robot post-training. By explicitly encoding the impact of robot capabilities (e.g., speed and payload) on collective behavior, CASH enables zero-shot generalization to unseen robots or team compositions. Our experiments involve multiple heterogeneous tasks, three learning paradigms (imitation learning, value-based, and policy-gradient RL), and SOTA multi-robot simulation (JaxMARL) and hardware (Robotarium) platforms. Across all conditions, we find that CASH generates appropriately-diverse behaviors and consistently outperforms baseline architectures in terms of performance and sample efficiency during both training and zero-shot generalization, all with 60%-80% fewer learnable parameters.

Glen Neville, Andrew Messing, Harish Ravichandar et al.

To realize effective heterogeneous multi-robot teams, researchers must leverage individual robots' relative strengths and coordinate their individual behaviors. Specifically, heterogeneous multi-robot systems must answer three important questions: \textit{who} (task allocation), \textit{when} (scheduling), and \textit{how} (motion planning). While specific variants of each of these problems are known to be NP-Hard, their interdependence only exacerbates the challenges involved in solving them together. In this paper, we present a novel framework that interleaves task allocation, scheduling, and motion planning. We introduce a search-based approach for trait-based time-extended task allocation named Incremental Task Allocation Graph Search (ITAGS). In contrast to approaches that solve the three problems in sequence, ITAGS's interleaved approach enables efficient search for allocations while simultaneously satisfying scheduling constraints and accounting for the time taken to execute motion plans. To enable effective interleaving, we develop a convex combination of two search heuristics that optimizes the satisfaction of task requirements as well as the makespan of the associated schedule. We demonstrate the efficacy of ITAGS using detailed ablation studies and comparisons against two state-of-the-art algorithms in a simulated emergency response domain.

Max Rudolph, Sonia Chernova, Harish Ravichandar

Multi-robot task allocation (MRTA) problems involve optimizing the allocation of robots to tasks. MRTA problems are known to be challenging when tasks require multiple robots and the team is composed of heterogeneous robots. These challenges are further exacerbated when we need to account for uncertainties encountered in the real-world. In this work, we address coalition formation in heterogeneous multi-robot teams with uncertain capabilities. We specifically focus on tasks that require coalitions to collectively satisfy certain minimum requirements. Existing approaches to uncertainty-aware task allocation either maximize expected pay-off (risk-neutral approaches) or improve worst-case or near-worst-case outcomes (risk-averse approaches). Within the context of our problem, we demonstrate the inherent limitations of unilaterally ignoring or avoiding risk and show that these approaches can in fact reduce the probability of satisfying task requirements. Inspired by models that explain foraging behaviors in animals, we develop a risk-adaptive approach to task allocation. Our approach adaptively switches between risk-averse and risk-seeking behavior in order to maximize the probability of satisfying task requirements. Comprehensive numerical experiments conclusively demonstrate that our risk-adaptive approach outperforms risk-neutral and risk-averse approaches. We also demonstrate the effectiveness of our approach using a simulated multi-robot emergency response scenario.

Abhinav Jain, Daphne Chen, Dhruva Bansal et al.

We address the problem of adapting robot trajectories to improve safety, comfort, and efficiency in human-robot collaborative tasks. To this end, we propose CoMOTO, a trajectory optimization framework that utilizes stochastic motion prediction models to anticipate the human's motion and adapt the robot's joint trajectory accordingly. We design a multi-objective cost function that simultaneously optimizes for i) separation distance, ii) visibility of the end-effector, iii) legibility, iv) efficiency, and v) smoothness. We evaluate CoMOTO against three existing methods for robot trajectory generation when in close proximity to humans. Our experimental results indicate that our approach consistently outperforms existing methods over a combined set of safety, comfort, and efficiency metrics.

Siddhartha Banerjee, Angel Daruna, David Kent et al.

Robot task execution when situated in real-world environments is fragile. As such, robot architectures must rely on robust error recovery, adding non-trivial complexity to highly-complex robot systems. To handle this complexity in development, we introduce Recovery-Driven Development (RDD), an iterative task scripting process that facilitates rapid task and recovery development by leveraging hierarchical specification, separation of nominal task and recovery development, and situated testing. We validate our approach with our challenge-winning mobile manipulator software architecture developed using RDD for the FetchIt! Challenge at the IEEE 2019 International Conference on Robotics and Automation. We attribute the success of our system to the level of robustness achieved using RDD, and conclude with lessons learned for developing such systems.

Pietro Pierpaoli, Harish Ravichandar, Nicholas Waytowich et al.

We present a technique for learning how to solve a multi-robot mission that requires interaction with an external environment by observing an expert system executing the same mission. We define the expert system as a team of robots equipped with a library of controllers, each designed to solve a specific task, supervised by an expert policy that appropriately selects controllers based on the states of robots and environment. The objective is for an un-trained team of robots (i.e., imitator system) equipped with the same library of controllers, but agnostic to the expert policy, to execute the mission, with performances comparable to those of the expert system. From un-annotated observations of the expert system, a multi-hypothesis filtering technique is used to estimate individual controllers executed by the expert policy. Then, the history of estimated controllers and environmental states is used to train a neural network policy for the imitator system. Considering a perimeter protection scenario on a team of differential-drive robots, we show that the learned policy endows the imitator system with performances comparable to those of the expert system.

Harish Ravichandar, S. Reza Ahmadzadeh, M. Asif Rana et al.

We propose a learning framework, named Multi-Coordinate Cost Balancing (MCCB), to address the problem of acquiring point-to-point movement skills from demonstrations. MCCB encodes demonstrations simultaneously in multiple differential coordinates that specify local geometric properties. MCCB generates reproductions by solving a convex optimization problem with a multi-coordinate cost function and linear constraints on the reproductions, such as initial, target, and via points. Further, since the relative importance of each coordinate system in the cost function might be unknown for a given skill, MCCB learns optimal weighting factors that balance the cost function. We demonstrate the effectiveness of MCCB via detailed experiments conducted on one handwriting dataset and three complex skill datasets.

Harish Ravichandar, Kenneth Shaw, Sonia Chernova

Large teams of heterogeneous agents have the potential to solve complex multi-task problems that are intractable for a single agent working independently. However, solving complex multi-task problems requires leveraging the relative strengths of the different kinds of agents in the team. We present Stochastic TRAit-based Task Assignment (STRATA), a unified framework that models large teams of heterogeneous agents and performs effective task assignments. Specifically, given information on which traits (capabilities) are required for various tasks, STRATA computes the assignments of agents to tasks such that the trait requirements are achieved. Inspired by prior work in robot swarms and biodiversity, we categorize agents into different species (groups) based on their traits. We model each trait as a continuous variable and differentiate between traits that can and cannot be aggregated from different agents. STRATA is capable of reasoning about both species-level and agent-level variability in traits. Further, we define measures of diversity for any given team based on the team's continuous-space trait model. We illustrate the necessity and effectiveness of STRATA using detailed experiments based in simulation and in a capture-the-flag game environment.