Nine SLMC papers accepted at ICRA 2021
30 May - 5 June 2021 Xi'an, China
Nine SLMC papers (three with IEEE RA-L option) have been accepted at the International Conference on Robotics and Automation (ICRA 2021) to be held in Xi'an, China. [Note: This page will be updated with camera ready versions of the papers and videos as they become available]
Decentralized Ability-Aware Adaptive Control for Multi-robot Collaborative Manipulation
Lei Yan, Theodorous Stouraitis and Sethu Vijayakumar, IEEE Robotics and Automation Letters (RAL), 2021 [pdf] [DOI] [video]
Presented at: IEEE International Conference on Robotics and Automation (ICRA 2021), Xian, China (2021)
Multi-robot collaboration is extremely challenging due to the different kinematic and dynamics capabilities of the robots, the limited communication between them, and the uncertainty of the system parameters. To address these challenges, we propose a Decentralized Ability-Aware Adaptive Control (DA3C) method, in which the force capability of each robot is maximized by exploiting its null-space motion, while the designed adaptive controller enables decentralized coordination according to the capability of each robot. Simulation results show the proposed method can achieve online adaptation and accurate trajectory tracking irrespective of the low-level controllers, and can be used for heterogeneous multi-robot systems.
RigidFusion: Robot Localisation and Mapping in Environments with Large Dynamic Rigid Objects
Ran Long, Christian Rauch, Tianwei Zhang, Vladimir Ivan and Sethu Vijayakumar, IEEE Robotics and Automation Letters (RAL), 2021 [pdf] [DOI] [video]
Presented at: IEEE International Conference on Robotics and Automation (ICRA 2021), Xian, China (2021)
RigidFusion is a state-of-the-art dense SLAM method that is robust to large dynamic occlusion (over 65%) in the scene, without requiring prior shape or appearance of the dynamic objects. It also contributes a pipeline to simultaneously segment, track and reconstruct the static background and one dynamic rigid body from RGB-D sequences. Importantly, we publish the dataset with the camera and object ground truth trajectories for benchmarking future work in the area of SLAM in dynamic environments.
Robust Footstep Planning and LQR Control for Dynamic Quadrupedal Locomotion
Guiyang Xin, Songyan Xin, Oguzhan Cebe, Mathew Jose Pollayil, Franco Angelini, Manolo Garabini, Sethu Vijayakumar and Michael Mistry, IEEE Robotics and Automation Letters (RAL), 2021 [pdf] [DOI] [video]
Presented at: IEEE International Conference on Robotics and Automation (ICRA 2021), Xian, China (2021)
We improve the robustness of dynamic quadrupedal locomotion based on two contributions: 1) fast model predictive foothold planning, and 2) application of LQR to projected inverse dynamic control for robust trajectory tracking. Experiments on the quadruped ANYmal demonstrate the effectiveness of the proposed method to react to external disturbances and environmental uncertainties.
Inverse Dynamics vs. Forward Dynamics in Direct Transcription Formulations for Trajectory Optimization
Henrique Ferrolho, Vladimir Ivan, Wolfgang Xaver Merkt, Ioannis Havoutis and Sethu Vijayakumar, Proc. IEEE International Conference on Robotics and Automation (ICRA 2021), Xian, China, 2021. [pdf] [video] [digest] [online presentation]
Direct transcription is a powerful technique that uses numerical optimisation to solve motion planning problems. Such numerical formulations use mathematical constraints to enforce motion requirements; in robotics, those constraints are used for, e.g., body placement, contact positions, system dynamics. We discuss two possible approaches to enforce nonlinear whole-body dynamics of robots in direct transcription: forward dynamics vs. inverse dynamics. Results show that using inverse dynamics is faster, requires less iterations, and is more robust to coarse problem discretisations.
Task-Space Decomposed Motion Planning Framework for Multi-Robot Loco-Manipulation
Xiaoyu Zhang, Lei Yan, Tin Lun Lam and Sethu Vijayakumar, Proc. IEEE International Conference on Robotics and Automation (ICRA 2021), Xian, China, 2021. [pdf] [video] [digest] [online presentation]
When several manipulators hold an object, closed-chain kinematic constraints are formed, and it will make the motion planning problems challenging by inducing lower-dimensional singularities. We propose a novel task-space decomposed motion planning framework for multi-robot simultaneous locomotion and manipulation. By decomposing the constrained task-space into different convex regions in a global planner, the local planner can efficiently generate a valid configuration for a multi-mobile-manipulator system. We demonstrate the proposed method in several simulations, where the robot team transports the object toward the goal in the obstacle-rich environments.
Sparsity-Inducing Optimal Control via Differential Dynamic Programming
Traiko Dinev, Wolfgang Xaver Merkt, Vladimir Ivan, Ioannis Havoutis and Sethu Vijayakumar, Proc. IEEE International Conference on Robotics and Automation (ICRA 2021), Xian, China, 2021. [pdf] [video] [digest] [online presentation]
Our paper describes how to use sparse controls in dynamic motion planning. We apply sparsity-inducing costs to plan satellite maneuvers, where thrusters using liquid propellants can only be switched on and off and can not provide variable thrust. We also apply sparsity in controls to a humanoid reaching task, which allows us to select the required number of joints for this lower-dimensional motion. We analyze the properties of a family of soft sparse costs and give insight into how to tune their free parameters.
Versatile Locomotion by Integrating Ankle, Hip, Stepping, and Height Variation Strategies
Jiatao Ding, Songyan Xin, Tin Lun Lam and Sethu Vijayakumar, Proc. IEEE International Conference on Robotics and Automation (ICRA 2021), Xian, China, 2021. [pdf] [video] [digest] [online presentation]
Bipedal and quadrupedal locomotion control on uneven terrain with height restrictions are a complex problem. We propose an enhanced Nonlinear Model Predictive Control (NMPC) approach for robust and adaptable walking. Multiple balancing strategies and their most effective combinations are explored including CoP modulation (if equipped with finite-sized feet), footstep adjustment, upper-body rotation, and vertical height variation.
A Passive Navigation Planning Algorithm for Collision-free Control of Mobile Robots
Carlo Tiseo, Vladimir Ivan, Wolfgang Xaver Merkt, Ioannis Havoutis, Michael Mistry and Sethu Vijayakumar, Proc. IEEE International Conference on Robotics and Automation (ICRA 2021), Xian, China, 2021. [pdf] [video] [digest] [online presentation]
A passive planning algorithm capable of autonomous obstacle avoidance in a domain with small concavity is presented. This novel method uses a passive controller that enables the navigation of complex dynamic maps without relying on numerical optimisation. Simulation and experimental results show that the technique can generate smooth, stable trajectories in drones and wheeled robots. The small computational cost enables scalability to swarm applications where the agents' movements are synchronised by issuing coordinated targets.
Whole Body Model Predictive Control with Memory of Motion: Experiments on a Torque-Controlled TALOS
Ewen Dantec, Rohan Budhiraja, Adria Roig, Teguh Santoso Lembono, Guilhem Saurel, Olivier Stasse, Pierre Fernbach, Steve Tonneau, Sylvain Calinon, Sethu Vijayakumar, Michel Taix and Nicolas Mansard, Proc. IEEE International Conference on Robotics and Automation (ICRA 2021), Xian, China (2021). [pdf] [video] [digest] [online presentation]
We present the first successful experiment implementing whole-body model predictive control with state feedback on a torque-control humanoid robot. We demonstrate that our control scheme is able to do whole-body target tracking, control the balance in front of strong external perturbations and avoid collision with an external object. The key elements for this success are threefold. First, optimal control over a receding horizon is implemented with Crocoddyl, an optimal control library based on differential dynamics programming, providing state-feedback control in less than 10 msecs. Second, a warm start strategy based on memory of motion has been implemented to overcome the sensitivity of the optimal control solver to initial conditions. Finally, the optimal trajectories are executed by a low-level torque controller, feedbacking on direct torque measurement at high frequency.