Whole-Body Nonlinear Model Predictive Control Through Contacts for Quadrupeds
arXiv:1712.02889 · doi:10.1109/LRA.2018.2800124
Abstract
In this work we present a whole-body Nonlinear Model Predictive Control approach for Rigid Body Systems subject to contacts. We use a full dynamic system model which also includes explicit contact dynamics. Therefore, contact locations, sequences and timings are not prespecified but optimized by the solver. Yet, thorough numerical and software engineering allows for running the nonlinear Optimal Control solver at rates up to 190 Hz on a quadruped for a time horizon of half a second. This outperforms the state of the art by at least one order of magnitude. Hardware experiments in form of periodic and non-periodic tasks are applied to two quadrupeds with different actuation systems. The obtained results underline the performance, transferability and robustness of the approach.
arXiv commentSubmitted to "Robotics and Automation: Letters" / "International Conference on Robotics and Automation 2018"
References in corpus (8)
- An Efficient Optimal Planning and Control Framework For Quadrupedal Locomotion
- Planning and Execution of Dynamic Whole-Body Locomotion for a Hydraulic Quadruped on Challenging Terrain
- Trajectory generation for multi-contact momentum-control
- The Control Toolbox - An Open-Source C++ Library for Robotics, Optimal and Model Predictive Control
- Evaluating direct transcription and nonlinear optimization methods for robot motion planning
- Automatic Differentiation of Rigid Body Dynamics for Optimal Control and Estimation
- Hierarchical Planning of Dynamic Movements without Scheduled Contact Sequences
- Projection based whole body motion planning for legged robots
Cited by in corpus (42)
- Reaching the Limit in Autonomous Racing: Optimal Control versus Reinforcement Learning
- Highly Dynamic Quadruped Locomotion via Whole-Body Impulse Control and Model Predictive Control
- Multi-expert learning of adaptive legged locomotion
- Representation-Free Model Predictive Control for Dynamic Motions in Quadrupeds
- TAMOLS: Terrain-Aware Motion Optimization for Legged Systems
- Differential Dynamic Programming for Multi-Phase Rigid Contact Dynamics
- Versatile Multi-Contact Planning and Control for Legged Loco-Manipulation
- The Control Toolbox - An Open-Source C++ Library for Robotics, Optimal and Model Predictive Control
- Model Hierarchy Predictive Control of Robotic Systems
- Tailoring Solution Accuracy for Fast Whole-body Model Predictive Control of Legged Robots
- MPC-Net: A First Principles Guided Policy Search
- Hybrid iLQR Model Predictive Control for Contact Implicit Stabilization on Legged Robots
- Contact-Implicit Trajectory Optimization Based on a Variable Smooth Contact Model and Successive Convexification
- Data Efficient Reinforcement Learning for Legged Robots
- Contact-Implicit Model Predictive Control: Controlling Diverse Quadruped Motions Without Pre-Planned Contact Modes or Trajectories
- Efficient Analytical Derivatives of Rigid-Body Dynamics using Spatial Vector Algebra
- Accelerating Second-Order Differential Dynamic Programming for Rigid-Body Systems
- Whole-body MPC for highly redundant legged manipulators: experimental evaluation with a 37 DoF dual-arm quadruped
- CACTO: Continuous Actor-Critic with Trajectory Optimization -- Towards global optimality
- Combining Coarse and Fine Physics for Manipulation using Parallel-in-Time Integration
- Adaptive Complexity Model Predictive Control
- Parameterized and GPU-Parallelized Real-Time Model Predictive Control for High Degree of Freedom Robots
- Efficient solution method based on inverse dynamics for optimal control problems of rigid body systems
- Towards General and Autonomous Learning of Core Skills: A Case Study in Locomotion
- Model Predictive Control for Trajectory Tracking on Differentiable Manifolds
- On the Benefits of GPU Sample-Based Stochastic Predictive Controllers for Legged Locomotion
- Parareal with a Learned Coarse Model for Robotic Manipulation
- Learning a Structured Neural Network Policy for a Hopping Task
- Practical Reinforcement Learning For MPC: Learning from sparse objectives in under an hour on a real robot
- ADDOPT: An Additive Manufacturing Optimal Control Framework Demonstrated in Minimizing Layer-Level Thermal Variance in Electron Beam Powder Bed Fusion
- Non-Gaited Legged Locomotion with Monte-Carlo Tree Search and Supervised Learning
- Full-Body Torque-Level Non-linear Model Predictive Control for Aerial Manipulation
- Follow the Gradient: Crossing the Reality Gap using Differentiable Physics (RealityGrad)
- Efficient Riccati recursion for optimal control problems with pure-state equality constraints
- Model Predictive Control with Environment Adaptation for Legged Locomotion
- Hybrid Systems Differential Dynamic Programming for Whole-Body Motion Planning of Legged Robots
- Probabilistic approach to feedback control enhances multi-legged locomotion on rugged landscapes
- Online Trajectory Optimization for Dynamic Aerial Motions of a Quadruped Robot
- How does the structure embedded in learning policy affect learning quadruped locomotion?
- An Efficient Paradigm for Feasibility Guarantees in Legged Locomotion
- Quadruped Locomotion on Non-Rigid Terrain using Reinforcement Learning
- Torque-Minimizing Control Allocation for Overactuated Quadrupedal Locomotion