Humanoid Robot Whole-Body Control With Phantom Inverse Kinematics
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional whole-body control systems for humanoid robots struggle to efficiently integrate task-space objectives with safety-related limitations in real time, primarily due to the complexity of inverse kinematics and inverse dynamics, which involves dozens of simultaneous variables and constraints.
Innovation Solution
A method that formulates a combined inverse kinematics problem as a quadratic program (QP) incorporating task, phantom inverse kinematics, and safety-related limits, using a whole-body controller to generate target joint parameters that satisfy both task and safety constraints, and a computing architecture that includes a movement controller, behavior manager, and model predictive control engine to optimize robot movements.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If conventional whole-body control systems are used to solve inverse kinematics and inverse dynamics problems, then the robot can perform tasks with multiple objectives, but the computational complexity increases and real-time performance deteriorates
Solution Approach 1:
The control system is segmented into multiple independent solvers: a first inverse kinematics solver that handles task-space objectives, and a second inverse kinematics solver that handles safety constraints and pose regularization. This segmentation allows each solver to focus on specific aspects of control, reducing the overall computational complexity while maintaining the ability to handle multiple objectives simultaneously.
Solution Approach 2:
A phantom inverse kinematics task is introduced as an intermediary construct that bridges the gap between task objectives and safety constraints. This phantom task is mathematically derived from the original task and incorporates safety-related limits, allowing the system to integrate multiple objectives through a unified mathematical framework without directly solving a complex multi-constraint optimization problem.
2Productivity
If conventional whole-body control systems solve inverse kinematics and inverse dynamics in real time, then task execution is achieved, but computational power consumption increases and battery life decreases
Solution Approach 1:
The computational workload is segmented and distributed across two specialized inverse kinematics solvers rather than using a single comprehensive solver. This division allows each solver to operate more efficiently on its specific subset of problems, reducing the overall computational power required while maintaining real-time task execution capability.
Solution Approach 2:
The phantom inverse kinematics task is pre-computed and integrated into the control framework before real-time execution. By preparing the mathematical formulation of safety constraints and pose regularization in advance, the system reduces the computational burden during real-time operation, thereby lowering power consumption while maintaining productivity.
3Reliability
If conventional whole-body control systems are used, then task execution is achieved, but safety constraints and pose regularization are difficult to integrate simultaneously
Solution Approach 1:
The phantom inverse kinematics task serves as an intermediary that naturally incorporates safety constraints and pose regularization into the control framework. By formulating these safety requirements as a mathematically derived phantom task, the system integrates multiple safety objectives through a unified mathematical framework, simplifying the control architecture while ensuring reliable safety constraint satisfaction.
Solution Approach 2:
The phantom inverse kinematics task is designed to be universal, simultaneously handling multiple safety-related objectives including joint limit constraints, collision avoidance, and pose regularization. This multi-functional approach allows a single mathematical construct to address diverse safety requirements, reducing the complexity of integrating multiple separate safety mechanisms.
Data Source
AI summary
The present disclosure provides a method for controlling a humanoid robot. The method comprises identifying a task for the humanoid robot to perform and establishing a safety-related limit. The method includes constructing a phantom inverse kinematics task mathematically derived from the identified task to regularize the robot's pose. The method formulates a combined inverse kinematics problem incorporating the task, phantom inverse kinematics task, and safety-related limit. The method solves the inverse kinematics problem to generate target joint parameters that satisfy the task while remaining within the safety-related limit. The method provides the target joint parameters to actuators to effect physical movement of the humanoid robot.