Robotics (cs.RO)

Abstract: In this paper, we review the question of which action space is best suited for controlling a real biped robot in combination with Sim2Real training. Position control has been popular as it has been shown to be more sample efficient and intuitive to combine with other planning algorithms. However, for position control gain tuning is required to achieve the best possible policy performance. We show that instead, using a torque-based action space enables task-and-robot agnostic learning with less parameter tuning and mitigates the sim-to-reality gap by taking advantage of torque control's inherent compliance. Also, we accelerate the torque-based-policy training process by pre-training the policy to remain upright by compensating for gravity. The paper showcases the first successful sim-to-real transfer of a torque-based deep reinforcement learning policy on a real human-sized biped robot. The video is available at https://youtu.be/CR6pTS39VRE.

Abstract: Our goal is to develop an efficient contact detection algorithm for large-scale GPU-based simulation of non-convex objects. Current GPU-based simulators such as IsaacGym and Brax must trade-off speed with fidelity, generality, or both when simulating non-convex objects. Their main issue lies in contact detection (CD): existing CD algorithms, such as Gilbert-Johnson-Keerthi (GJK), must trade off their computational speed with accuracy which becomes expensive as the number of collisions among non-convex objects increases. We propose a data-driven approach for CD, whose accuracy depends only on the quality and quantity of offline dataset rather than online computation time. Unlike GJK, our method inherently has a uniform computational flow, which facilitates efficient GPU usage based on advanced compilers such as XLA (Accelerated Linear Algebra). Further, we offer a data-efficient solution by learning the patterns of colliding local crop object shapes, rather than global object shapes which are harder to learn. We demonstrate our approach improves the efficiency of existing CD methods by a factor of 5-10 for non-convex objects with comparable accuracy. Using the previous work on contact resolution for a neural-network-based contact detector, we integrate our CD algorithm into the open-source GPU-based simulator, Brax, and show that we can improve the efficiency over IsaacGym and generality over standard Brax. We highly recommend the videos of our simulator included in the supplementary materials.

Abstract: There are many industrial, commercial and social applications for multi-agent planning for multirotors such as autonomous agriculture, infrastructure inspection and search and rescue. Thus, improving on the state-of-the-art of multi-agent planning to make it a viable real-world solution is of great benefit. In this work, we propose a new method for multi-agent planning in a static environment that improves our previous work by making it fully online as well as robust to communication latency. The proposed framework generates a global path and a Safe Corridor to avoid static obstacles in an online fashion (generated offline in our previous work). It then generates a time-aware Safe Corridor which takes into account the future positions of other agents to avoid intra-agent collisions. The time-aware Safe Corridor is given with a local reference trajectory to an MIQP (Mixed-Integer Quadratic Problem)/MPC (Model Predictive Control) solver that outputs a safe and optimal trajectory. The planning frequency is adapted to account for communication delays. The proposed method is fully online, real-time, decentralized, and synchronous. It is compared to 3 recent state-of-the-art methods in simulations. It outperforms all methods in robustness and safety as well as flight time. It also outperforms the only other state-of-the-art latency robust method in computation time.

Abstract: Dexterous in-hand manipulation for a multi-fingered anthropomorphic hand is extremely difficult because of the high-dimensional state and action spaces, rich contact patterns between the fingers and objects. Even though deep reinforcement learning has made moderate progress and demonstrated its strong potential for manipulation, it is still faced with certain challenges, such as large-scale data collection and high sample complexity. Especially, for some slight change scenes, it always needs to re-collect vast amounts of data and carry out numerous iterations of fine-tuning. Remarkably, humans can quickly transfer learned manipulation skills to different scenarios with little supervision. Inspired by human flexible transfer learning capability, we propose a novel dexterous in-hand manipulation progressive transfer learning framework (PTL) based on efficiently utilizing the collected trajectories and the source-trained dynamics model. This framework adopts progressive neural networks for dynamics model transfer learning on samples selected by a new samples selection method based on dynamics properties, rewards and scores of the trajectories. Experimental results on contact-rich anthropomorphic hand manipulation tasks show that our method can efficiently and effectively learn in-hand manipulation skills with a few online attempts and adjustment learning under the new scene. Compared to learning from scratch, our method can reduce training time costs by 95%.

Abstract: This paper provides an overview of the current state-of-the-art in selective harvesting robots (SHRs) and their potential for addressing the challenges of global food production. SHRs have the potential to increase productivity, reduce labour costs, and minimise food waste by selectively harvesting only ripe fruits and vegetables. The paper discusses the main components of SHRs, including perception, grasping, cutting, motion planning, and control. It also highlights the challenges in developing SHR technologies, particularly in the areas of robot design, motion planning and control. The paper also discusses the potential benefits of integrating AI and soft robots and data-driven methods to enhance the performance and robustness of SHR systems. Finally, the paper identifies several open research questions in the field and highlights the need for further research and development efforts to advance SHR technologies to meet the challenges of global food production. Overall, this paper provides a starting point for researchers and practitioners interested in developing SHRs and highlights the need for more research in this field.

Abstract: In this work we propose a coverage planning control approach which allows a mobile agent, equipped with a controllable sensor (i.e., a camera) with limited sensing domain (i.e., finite sensing range and angle of view), to cover the surface area of an object of interest. The proposed approach integrates ray-tracing into the coverage planning process, thus allowing the agent to identify which parts of the scene are visible at any point in time. The problem of integrated ray-tracing and coverage planning control is first formulated as a constrained optimal control problem (OCP), which aims at determining the agent's optimal control inputs over a finite planning horizon, that minimize the coverage time. Efficiently solving the resulting OCP is however very challenging due to non-convex and non-linear visibility constraints. To overcome this limitation, the problem is converted into a Markov decision process (MDP) which is then solved using reinforcement learning. In particular, we show that a controller which follows an optimal control law can be learned using off-policy temporal-difference control (i.e., Q-learning). Extensive numerical experiments demonstrate the effectiveness of the proposed approach for various configurations of the agent and the object of interest.

Abstract: Robot-assisted intervention has shown reduced radiation exposure to physicians and improved precision in clinical trials. However, existing vascular robotic systems follow master-slave control mode and entirely rely on manual commands. This paper proposes a novel offline reinforcement learning algorithm, Conservative Actor-critic with SmOoth Gradient (CASOG), to learn manipulation skills from human demonstrations on vascular robotic systems. The proposed algorithm conservatively estimates Q-function and smooths gradients of convolution layers to deal with distribution shift and overfitting issues. Furthermore, to focus on complex manipulations, transitions with larger temporal-difference error are sampled with higher probability. Comparative experiments in a pre-clinical environment demonstrate that CASOG can deliver guidewire to the target at a success rate of 94.00\% and mean backward steps of 14.07, performing closer to humans and better than prior offline reinforcement learning methods. These results indicate that the proposed algorithm is promising to improve the autonomy of vascular robotic systems.

Abstract: This work contributes to developing an agent based on deep reinforcement learning capable of acting in a beyond visual range (BVR) air combat simulation environment. The paper presents an overview of building an agent representing a high-performance fighter aircraft that can learn and improve its role in BVR combat over time based on rewards calculated using operational metrics. Also, through self-play experiments, it expects to generate new air combat tactics never seen before. Finally, we hope to examine a real pilot's ability, using virtual simulation, to interact in the same environment with the trained agent and compare their performances. This research will contribute to the air combat training context by developing agents that can interact with real pilots to improve their performances in air defense missions.

Abstract: This paper proposes a method to regulate the tissue temperature during laser surgery by robotically controlling the laser focus. Laser-tissue interactions are generally considered hard to control due to the inherent inhomogeneity of biological tissue, which can create significant variability in its thermal response to laser irradiation. In this study, we use methods from nonlinear control theory to synthesize a temperature controller capable of working on virtually any tissue type without any prior knowledge of its physical properties. The performance of the controller is evaluated in ex-vivo experiments.

Abstract: Task and motion planning is one of the key problems in robotics today. It is often formulated as a discrete task allocation problem combined with continuous motion planning. Many existing approaches to TAMP involve explicit descriptions of task primitives that cause discrete changes in the kinematic relationship between the actor and the objects. In this work we propose an alternative approach to TAMP which does not involve explicit enumeration of task primitives. Instead, the actions are represented implicitly as part of the solution to a nonlinear optimization problem. We focus on decision making for robotic manipulators, specifically for pick and place tasks, and show several possible extensions. We explore the efficacy of the model through a number of simulated experiments involving multiple robots, objects and interactions with the environment.

Abstract: The research on multi-robot coverage path planning (CPP) has been attracting more and more attention. In order to achieve efficient coverage, this paper proposes an improved DARP coverage algorithm. The improved DARP algorithm based on A* algorithm is used to assign tasks to robots and then combined with STC algorithm based on Up-First algorithm to achieve full coverage of the task area. Compared with the initial DARP algorithm, this algorithm has higher efficiency and higher coverage rate.

Abstract: We present a system that enables an autonomous small-scale RC car to drive aggressively from visual observations using reinforcement learning (RL). Our system, FastRLAP (faster lap), trains autonomously in the real world, without human interventions, and without requiring any simulation or expert demonstrations. Our system integrates a number of important components to make this possible: we initialize the representations for the RL policy and value function from a large prior dataset of other robots navigating in other environments (at low speed), which provides a navigation-relevant representation. From here, a sample-efficient online RL method uses a single low-speed user-provided demonstration to determine the desired driving course, extracts a set of navigational checkpoints, and autonomously practices driving through these checkpoints, resetting automatically on collision or failure. Perhaps surprisingly, we find that with appropriate initialization and choice of algorithm, our system can learn to drive over a variety of racing courses with less than 20 minutes of online training. The resulting policies exhibit emergent aggressive driving skills, such as timing braking and acceleration around turns and avoiding areas which impede the robot's motion, approaching the performance of a human driver using a similar first-person interface over the course of training.

Abstract: Legged robots have enormous potential in their range of capabilities, from navigating unstructured terrains to high-speed running. However, designing robust controllers for highly agile dynamic motions remains a substantial challenge for roboticists. Reinforcement learning (RL) offers a promising data-driven approach for automatically training such controllers. However, exploration in these high-dimensional, underactuated systems remains a significant hurdle for enabling legged robots to learn performant, naturalistic, and versatile agility skills. We propose a framework for training complex robotic skills by transferring experience from existing controllers to jumpstart learning new tasks. To leverage controllers we can acquire in practice, we design this framework to be flexible in terms of their source -- that is, the controllers may have been optimized for a different objective under different dynamics, or may require different knowledge of the surroundings -- and thus may be highly suboptimal for the target task. We show that our method enables learning complex agile jumping behaviors, navigating to goal locations while walking on hind legs, and adapting to new environments. We also demonstrate that the agile behaviors learned in this way are graceful and safe enough to deploy in the real world.

Abstract: Learning-based control algorithms have led to major advances in robotics at the cost of decreased safety guarantees. Recently, neural networks have also been used to characterize safety through the use of barrier functions for complex nonlinear systems. Learned barrier functions approximately encode and enforce a desired safety constraint through a value function, but do not provide any formal guarantees. In this paper, we propose a local dynamic programming (DP) based approach to "patch" an almost-safe learned barrier at potentially unsafe points in the state space. This algorithm, HJ-Patch, obtains a novel barrier that provides formal safety guarantees, yet retains the global structure of the learned barrier. Our local DP based reachability algorithm, HJ-Patch, updates the barrier function "minimally" at points that both (a) neighbor the barrier safety boundary and (b) do not satisfy the safety condition. We view this as a key step to bridging the gap between learning-based barrier functions and Hamilton-Jacobi reachability analysis, providing a framework for further integration of these approaches. We demonstrate that for well-trained barriers we reduce the computational load by 2 orders of magnitude with respect to standard DP-based reachability, and demonstrate scalability to a 6-dimensional system, which is at the limit of standard DP-based reachability.