The field of robotics is witnessing significant advancements in legged locomotion and control, with a focus on adaptive and robust systems. Researchers are developing innovative algorithms and frameworks that enable quadrupedal robots to navigate complex and unknown environments, such as planetary surfaces and uneven terrains. These developments are crucial for future autonomous missions and applications. Noteworthy papers in this area include: An adaptive hierarchical control framework for quadrupedal robots, which combines model-based dynamic control with online model adaptation and adaptive footstep planning. A reinforcement learning-based robust wall climbing locomotion controller, which achieves a high success rate and strong adhesion retention in simulation and hardware experiments. Real-Time Gait Adaptation for Quadrupeds using Model Predictive Control and Reinforcement Learning, which proposes an optimization framework for real-time gait adaptation in a continuous gait space.
Advances in Legged Locomotion and Robot Control
Sources
Adaptive Legged Locomotion via Online Learning for Model Predictive Control
The Formalism-Implementation Gap in Reinforcement Learning Research
An adaptive hierarchical control framework for quadrupedal robots in planetary exploration
Towards Proprioceptive Terrain Mapping with Quadruped Robots for Exploration in Planetary Permanently Shadowed Regions
Motion Planning and Control of an Overactuated 4-Wheel Drive with Constrained Independent Steering
Safe Active Navigation and Exploration for Planetary Environments Using Proprioceptive Measurements
Calibration of Parallel Kinematic Machine Based on Stewart Platform-A Literature Review
Reinforcement Learning-based Robust Wall Climbing Locomotion Controller in Ferromagnetic Environment
Multi-Modal Decentralized Reinforcement Learning for Modular Reconfigurable Lunar Robots
Balancing Specialization and Centralization: A Multi-Agent Reinforcement Learning Benchmark for Sequential Industrial Control
Robot Path and Trajectory Planning Considering a Spatially Fixed TCP
A Parameter-Linear Formulation of the Optimal Path Following Problem for Robotic Manipulator
Real-Time Gait Adaptation for Quadrupeds using Model Predictive Control and Reinforcement Learning