Robot Trajectory Planning With Candidate Evaluation for Real-Time Motion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing trajectory generation systems for mobile bodies, such as robots, take a long time to compute trajectories, particularly when generating paths for grippers, leading to increased computation time without ensuring accuracy.
Innovation Solution
A trajectory generation system that includes a computing unit generating multiple trajectory candidates with degrees of freedom associated with grip conditions within predetermined time, processing, and number constraints, evaluated and checked for feasibility to select optimal trajectories, allowing for real-time movement while maintaining accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the trajectory generation system computes multiple trajectory candidates with high accuracy, then the trajectory accuracy is improved, but the computation time increases significantly
Solution Approach 1:
The patent divides the trajectory generation process into two distinct phases: a planning phase that generates multiple trajectory candidates with high accuracy, and an execution phase that selects and follows a simplified trajectory. This segmentation allows computationally intensive calculations to be performed only when needed for planning, while real-time execution uses pre-computed data, thereby resolving the contradiction between accuracy and computation time.
Solution Approach 2:
The system performs preliminary computation of multiple trajectory candidates in advance during the planning phase. By pre-computing and storing trajectory candidates with associated cost values before actual execution, the system eliminates the need for real-time computation during movement, thus maintaining high accuracy while reducing real-time computation time.
2Reliability
If the system generates a large number of trajectory candidates, then the trajectory quality is improved, but the processing amount increases
Solution Approach 1:
The patent generates a predetermined number of trajectory candidates (e.g., 10-100 candidates) which is sufficient to ensure high trajectory quality but not excessively large. This partial action approach generates enough candidates to find an optimal path while avoiding the computational burden of generating all possible trajectories, thus balancing reliability with processing complexity.
Solution Approach 2:
The system uses cost functions with multiple parameters (trajectory length, safety margins, energy consumption, time) to evaluate and rank trajectory candidates. By changing and weighting these parameters appropriately, the system can generate a manageable number of high-quality candidates rather than exhaustively computing all possible trajectories, thus improving trajectory quality without excessive processing.
3Speed
If the trajectory generation system ensures real-time movement, then the responsiveness is improved, but the computation time available for trajectory generation is reduced
Solution Approach 1:
The patent separates trajectory generation into offline planning (computation-intensive) and online execution (time-critical). During offline planning, multiple candidates are computed with full accuracy. During online execution, the system simply selects from pre-computed candidates and generates minimal refinement, ensuring real-time responsiveness without sacrificing overall trajectory quality.
Solution Approach 2:
The system performs preliminary trajectory candidate generation and evaluation before real-time execution begins. By pre-computing trajectory candidates and their cost values during planning phases, the system ensures that during real-time movement, only lightweight selection and tracking operations are needed, thus achieving both real-time responsiveness and computational accuracy.
Data Source
AI summary
A trajectory generation system includes a computing unit that generates a trajectory on which a mobile body or a gripper moves from a start position to a target position, and an evaluating unit that evaluates a plurality of trajectory candidates. The computing unit generates the trajectory candidates leading to a target area including the target position and its vicinity, with a plurality of degrees of freedom associated with predetermined grip conditions, within at least one of a range in which the candidates can be computed in a predetermined period, a range in which they can be computed in a predetermined processing amount, and a range in which a predetermined number of trajectory candidates can be computed. The evaluating unit conducts evaluation of the trajectory candidates generated, based on an evaluation item, and the computing unit generates the trajectory, using one trajectory candidate selected based on the evaluation.