Abstract:Safe path and gait planning are essential for bipedal robots to navigate complex real-world environments. The prevailing approaches often plan the path and gait separately in a hierarchical fashion, potentially resulting in unsafe movements due to neglecting the physical constraints of walking robots. A safety-critical path must not only avoid obstacles but also ensure that the robot's gaits are subject to its dynamic and kinematic constraints. This work presents a novel approach that unifies path planning and gait planning via a Model Predictive Control (MPC) using the Linear Inverted Pendulum (LIP) model representing bipedal locomotion. This approach considers environmental constraints, such as obstacles, and the robot's kinematics and dynamics constraints. By using discrete-time Control Barrier Functions for obstacle avoidance, our approach generates the next foot landing position, ensuring robust walking gaits and a safe navigation path within clustered environments. We validated our proposed approach in simulation using a Digit robot in 20 randomly created environments. The results demonstrate improved performance in terms of safety and robustness when compared to hierarchical path and gait planning frameworks.

What problem does this paper attempt to address?

The paper mainly aims to address the following issues: ### Core Issues Addressed by the Paper - **Unified Path and Gait Planning**: Proposes a new method to unify path planning and gait planning to improve the navigation safety of biped robots in complex real-world environments. ### Specific Problem Description - **Limitations of Existing Methods**: Current methods often perform path planning and gait planning separately. This layered approach may lead to unsafe movements as it ignores the impact of the physical constraints of walking robots. - **Requirements for a Safe Path**: A safe path not only needs to avoid obstacles but also ensure that the robot's gait complies with its dynamic and kinematic constraints. - **Challenges of Unified Planning**: How to achieve the unification of path planning and gait planning through Model Predictive Control (MPC) combined with the Linear Inverted Pendulum (LIP) model. ### Main Contributions - **MPC-based Unified Planning Framework**: Proposes a new framework based on MPC, using the LIP model to represent bipedal walking, and considering environmental constraints (such as obstacles), as well as the robot's kinematic and dynamic constraints. - **Application of Discrete-Time Control Barrier Functions**: Uses Discrete-Time Control Barrier Functions (D-CBFs) to ensure that the next foot placement position can avoid obstacles in dense environments, thereby ensuring robust walking gait and safe navigation paths. - **Performance Verification**: Verified using the Digit robot in simulations, tested in 20 randomly created environments, showing improvements in safety and robustness compared to traditional layered path and gait planning. ### Technical Details - **Extension of the LIP Model**: Introduces a 3D Linear Inverted Pendulum model that includes a steering angle state to better describe the robot's steering actions. - **Multi-Step Preview Plan**: Uses a multi-step preview plan of the LIP model to determine the next stepping position through the MPC formulation. - **Environmental and Kinematic Safety Constraints**: Constructs discrete control barrier functions to avoid collisions and incorporates the robot's kinematic constraints to ensure the safety and feasibility of the planning. ### Experimental Results - **Improved Performance**: By adding steering angle costs and mobility constraints, the robot can navigate more smoothly in dense environments, reducing unnecessary adjustments and improving navigation safety and reliability. - **Comparison with Traditional Methods**: Compared to traditional layered planning frameworks, the new method shows better safety in multiple randomly generated environments, successfully reaching the target without collisions or falls. ### Conclusion This study proposes a novel method that effectively addresses the issue of safe navigation for biped robots in complex environments. By unifying path planning and gait planning, it significantly improves the robot's navigation performance in dense obstacle environments.

Unified Path and Gait Planning for Safe Bipedal Robot Navigation

Real-Time Safe Bipedal Robot Navigation using Linear Discrete Control Barrier Functions

Nonlinear MPC-Based Control Framework for Quadruped Robots: Touch-Down in Complex Terrain

Efficient Motion Planning Based on Kinodynamic Model for Quadruped Robots Following Persons in Confined Spaces

Model Predictive Control for Motion Planning of Quadrupedal Locomotion.

Safety-critical Motion Planning for Collaborative Legged Loco-Manipulation over Discrete Terrain

Terrain-Adaptive, ALIP-Based Bipedal Locomotion Controller via Model Predictive Control and Virtual Constraints

Contingency Model Predictive Control for Bipedal Locomotion on Moving Surfaces with a Linear Inverted Pendulum Model

Autonomous Navigation of Underactuated Bipedal Robots in Height-Constrained Environments

Bipedal Safe Navigation over Uncertain Rough Terrain: Unifying Terrain Mapping and Locomotion Stability

Multi-Layered Safety for Legged Robots via Control Barrier Functions and Model Predictive Control

Safety-critical Locomotion of Biped Robots in Infeasible Paths: Overcoming Obstacles during Navigation toward Destination

Socially Acceptable Bipedal Navigation: A Signal-Temporal-Logic- Driven Approach for Safe Locomotion

Integrated Task and Motion Planning for Safe Legged Navigation in Partially Observable Environments

Global Position Control on Underactuated Bipedal Robots: Step-to-step Dynamics Approximation for Step Planning

Towards Safe Locomotion Navigation in Partially Observable Environments with Uneven Terrain

Robot Safe Planning In Dynamic Environments Based On Model Predictive Control Using Control Barrier Function

Bipedal Walking on Constrained Footholds with MPC Footstep Control

Perceptive Locomotion of Legged Robot Coupling Model Predictive Control and Terrain Mapping

Real-Time Control of Full Actuated Biped Robot Based on Nonlinear Model Predictive Control

Blended CPG and Independent Foot Trajectory Planning for Quadruped Robot.