Trajectory Generation for Humanoid Robots
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing trajectory generation systems for humanoid robots can result in discrete temporal differentiation values, such as velocity or acceleration, when the specified time for foot contact with the ground is varied, leading to non-smooth trajectories.
Innovation Solution
A trajectory generation system comprising a first element that sets variable times and a second element that generates a function over a specified period, ensuring the integral value of the third-order temporal differentiation function is minimal, maintaining smoothness of first and second-order temporal differentiation functions while allowing time variations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If the specified time for foot contact is varied to adapt to different walking conditions, then the adaptability of the trajectory generation system is improved, but the smoothness of temporal differentiation functions (velocity, acceleration trajectories) deteriorates and becomes discrete
Solution Approach 1:
The patent changes the parameter being optimized from the traditional approach (matching position at specified time) to minimizing the integral of the absolute value of the third-order temporal differentiation function. This parameter change allows the system to maintain smoothness in velocity and acceleration trajectories even when the specified time for foot contact is varied, thereby resolving the contradiction between adaptability and smoothness.
2Manufacturing precision
If the trajectory is generated to match the desired value at the specified time using a dead beat filter, then the positioning accuracy is improved, but the temporal differentiation trajectories become discrete when specified time varies
Solution Approach 1:
The patent changes the optimization parameter from position matching (dead beat filter approach) to minimizing the integral of the third-order temporal differentiation. This allows the system to achieve both positioning accuracy and smooth velocity/acceleration trajectories simultaneously, even when the specified time varies.
Solution Approach 2:
The patent introduces a dynamic optimization approach where the trajectory is generated by minimizing the integral of the absolute value of the third-order temporal differentiation function over the specified period. This dynamic approach adapts to varying specified times while maintaining smoothness in the temporal differentiation functions, unlike static dead beat filters that produce discrete trajectories when time varies.
Data Source
AI summary
A specified function p is generated to satisfy a first condition that an integral value of an evaluation function over a specified period is minimal. The evaluation function denotes a magnitude of an absolute value of a third order specified function p(3)(t). Moreover, the specified function p is generated to satisfy a second condition that a value of a second order specified function p(2)(t) is limited within a limiter range [p(2)min_c, p(2)max_c].


