APT-RL (action pretrained transformer-based reinforcement learning), a unified framework that enables multiskill locomotion to achieve high-speed traversal in complex environments through autonomous skill transitions using only onboard perception and computation, is presented.
Abstract
Enabling quadrupedal robots to traverse complex terrains, from rugged outdoor environments to urban landscapes, requires seamless integration of multiple motor skills, smooth transitions between gaits, and high-speed perceptive locomotion using only onboard sensors. We present APT-RL (action pretrained transformer-based reinforcement learning), a unified framework that enables multiskill locomotion to achieve high-speed traversal in complex environments through autonomous skill transitions using only onboard perception and computation. Our approach generates large-scale, feature-rich two-dimensional (2D) motion datasets through trajectory optimization with simplified dynamics. These datasets enable training of diverse, reusable locomotion skills that transfer effectively to a real quadruped robot operating on complex uneven terrains. The resulting high-quality skills serve as strong priors for efficient learning of complex downstream tasks and extend naturally to 3D environments, enabling smooth, high-speed multiskill locomotion in deployed policy. Real-world experiments demonstrate the framework's capabilities: The robot performed agile maneuvers through complex indoor obstacles and outdoor wild environments, including dynamic drop-down maneuvers that reached instantaneous peak speeds of up to 6 meters per second. A single onboard policy enabled robust traversal of diverse obstacles, including stairs, hurdles, stepping stones, gaps, and fallen branches, demonstrating the versatility and effectiveness of our approach.
This work proposes a hierarchical reinforcement learning pipeline that empowers the robots to perform aggressive locomotion through constrained obstacles--a narrow gate, extending the lifelike agility of legged robots to match that of their biological counterparts.
Zeren Luo, Jiahui Zhang, Yimin Han et al.· 1 citation
Quadruped robots have demonstrated impressive agility in parkour locomotion across complex terrains. However, most systems still rely on human intervention for high-level planning, and autonomous parkour navigation remains underexplored. The key challenges include fine-grained velocity regulation, long-horizon anticipatory behaviors, and tight coupling between perception and embodied execution. To address these challenges, we propose a Multimodal Diffusion Policy (MulDP) that integrates visual perception with robot proprioception and goal information to generate temporally coherent and anticipatory navigation velocity commands, tightly coupling perception with embodied control to enable robust autonomous navigation. To support the training of MulDP, we construct the first Quadruped Parkour Navigation Dataset (QPND), a multimodal dataset that encompasses diverse navigation behaviors and complex terrains. Extensive simulation and real-world experiments demonstrate that MulDP enables robust long-horizon autonomous navigation and effective traversal across complex terrains.
Kang-Mai Hu, Yue-Qi Zhang, Peng Zhai et al.· 0 citations
Autonomous mobile robotic platforms operating in unstructured environments face significant challenges due to unpredictable terrain topologies, varying soil properties, and geometric obstacles. Traditional reactive motion planners often fail or suffer severe efficiency losses when transitioning between highly distinct surfaces such as sand, gravel, mud, and solid rock. This paper introduces a comprehensive framework for intelligent terrain adaptation that combines exteroceptive perception and proprioceptive feedback to dynamically optimize robotic locomotion parameters. By leveraging deep learning-based visual-tactile classification alongside a real-time predictive control system, the proposed mobile robotic architecture achieves autonomous adaptation of wheel torque, suspension geometry, and path selection. Experimental validations conducted across four distinct simulated and real-world testing grounds demonstrate substantial improvements in energy efficiency, slip reduction, and stability control compared to traditional static locomotion algorithms. The results show that multi-modal sensor fusion provides the reliable situational awareness necessary for long-term robot autonomy in search-and-rescue, planetary exploration, and agricultural operations.
Rajesh Sharma, Priya Natarajan· International Journal of Int...· 0 citations
Traversing sparse 3D structures requires humanoid robots to perceive thin, overhanging geometry while executing agile, accurate whole-body motions. We study this problem through monkey-bar traversal, where the robot must jump to the structure, traverse it through sparse bar interactions, and land safely. For this task, we present a reinforcement-learning-based perceptive control system that operates directly on observations from a head-mounted solid-state lidar. To extract task-relevant geometry from the sparse returns, the policy consumes the raw lidar scan through an attention-based encoder with recurrent memory. This policy is obtained by a phase-scheduled teacher- student pipeline that combines privileged experts for jumping up, brachiating, and jumping down. For transfer to hardware, we model lidar noise, battery-voltage sag, and actuator thermal limits, and equip the humanoid with passive hook end-effectors for robust bar interaction. On hardware, the resulting policy completes the full jump-up->brachiation->jump-down sequence in 14 of 15 trials across three bar configurations and reaches brachiation speeds up to 0.5 m/s. Beyond brachiation, the same perception backbone supports a separately trained policy that ducks beneath thin overhead obstacles with 2 cm cross-sections.
Efe Ongan, Chong Zhang, Boyang Sun et al.· 0 citations
An integrated framework that combines a vision module for landing point and time prediction with a direct position and time conditioned RL locomotion policy, instead of intermediate velocity commands is proposed, which mitigates perception latency during dynamic interception.
Yi-Dong Zhu, Zibo Dai, Tong-Ning Zhang et al.· 0 citations
Robotic mobility remains a major challenge in planetary exploration, particularly in environments characterized by uncertain terrain interaction, low gravity, and irregular contact conditions. Conventional wheeled and gait-based locomotion strategies typically rely on predefined contact patterns and motion-centric control formulations, which can become fragile in highly unstructured extraterrestrial environments such as lava tubes, crater walls, and granular slopes. This perspective proposes a support-centric interpretation of locomotion in which mobility is viewed as the continuous generation, redistribution, and adaptation of support under uncertain interaction conditions. Rather than treating contact as a secondary constraint within trajectory execution, the proposed framework interprets locomotion through the evolution of support configurations, support quality, and contact reliability. The paper synthesizes developments in terramechanics, adaptive legged locomotion, bio-inspired robotics, and learning-based control to establish conceptual links between contact interaction and support evolution. A conceptual framework for learning support dynamics is further introduced to outline possible directions for adaptive multi-contact locomotion in space robotics. The proposed perspective is intended not as a replacement for existing locomotion methods, but as a higher-level framework for guiding future research in robust terrain-adaptive robotic mobility.
E. Prisăcariu, Oana Dumitrescu· Robotics· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
