Reference-Free Sampling-Based Model Predictive Control

2025-11-24

AI Summary

Achieving diverse dynamic locomotion for quadrupedal and humanoid robots remains challenging under conditions of no predefined gait patterns, no reference trajectories, and no offline training. Method: This paper proposes a sampling-based model predictive control (MPC) framework for autonomous emergence of locomotion behaviors. Its core innovation is a position-velocity dual-space B-spline parameterization, integrated with model predictive path integral (MPPI) control to enable online, adaptive optimization of foot contact and lift-off strategies using only a small number of sampled trajectories. Contribution/Results: The method eliminates reliance on manually designed contact sequences or reference trajectories, enabling real-time, autonomous generation of complex behaviors—including walking, jumping, handstand balancing, and backflips. It achieves pure-CPU real-time control on the Go2 robot and successfully demonstrates high-dynamic maneuvers in simulation. By unifying motion planning and control within a single adaptive optimization loop, the framework significantly improves generalizability and autonomy in legged locomotion control.

Technology Category

• Intelligent Robots: Behavior Learning & Control
• Humans and AI: Human-Aware Planning and Behavior Prediction
• Search and Optimization: Sampling/Simulation-based Search

Application Category

• Responsible Web: Machine-in-the-loop, human agency and autonomy
• User Modeling, Personalization and Recommendation: User modeling and simulation for interactive and conversational systems
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions

Abstract

We present a sampling-based model predictive control (MPC) framework that enables emergent locomotion without relying on handcrafted gait patterns or predefined contact sequences. Our method discovers diverse motion patterns, ranging from trotting to galloping, robust standing policies, jumping, and handstand balancing, purely through the optimization of high-level objectives. Building on model predictive path integral (MPPI), we propose a dual-space spline parameterization that operates on position and velocity control points. Our approach enables contact-making and contact-breaking strategies that adapt automatically to task requirements, requiring only a limited number of sampled trajectories. This sample efficiency allows us to achieve real-time control on standard CPU hardware, eliminating the need for GPU acceleration typically required by other state-of-the-art MPPI methods. We validate our approach on the Go2 quadrupedal robot, demonstrating various emergent gaits and basic jumping capabilities. In simulation, we further showcase more complex behaviors, such as backflips, dynamic handstand balancing and locomotion on a Humanoid, all without requiring reference tracking or offline pre-training.

Problem

• Develops reference-free MPC for emergent locomotion without predefined patterns
• Enables adaptive contact strategies using efficient trajectory sampling
• Achieves real-time control on CPU without GPU acceleration requirements

Innovation

• Sampling-based MPC without predefined gait patterns
• Dual-space spline parameterization for position and velocity
• Real-time control on CPU with limited trajectory samples