Vehicle Trajectory Generation for Multi-Direction Obstacle Avoidance

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing trajectory generation devices face challenges in generating future trajectories for mobile bodies to effectively avoid interference objects, particularly when the trajectory requires bending in multiple directions, leading to increased response delays and reduced spatial margins for interference avoidance, especially in vehicles where feasible trajectories are limited.

Innovation Solution

A trajectory generation device that includes a surrounding situation recognizer, a predicted trajectory generator, an interference degree parameter calculator, and a correction amount determiner, which divides the future trajectory into multiple predicted trajectories using different trajectory models and calculates correction amounts to minimize interference, allowing for appropriate trajectory adjustment based on the surrounding situation and the position of the mobile body relative to the interference object.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If multiple extreme value search controllers are used to generate trajectories bending in multiple directions, then the trajectory can avoid interference objects, but the number of taps of moving average filter increases causing response delay

Engineering Contradiction:
Improvetrajectory bending capabilityVSAvoidresponse delay
Core Design Contradiction:
Adaptability or versatilityVSLoss of time

Solution Approach 1:

The patent merges multiple extreme value search controllers into a single controller that processes multiple evaluation functions simultaneously. This consolidation reduces the number of moving average filters needed, thereby decreasing the total number of taps and reducing response delay while maintaining the capability to generate multi-directional trajectories.

Inventive Principle:
Principle #5Merging (Combining)

Solution Approach 2:

The single extreme value search controller is designed to handle multiple evaluation functions (J1, J2, J3) that correspond to different trajectory bending directions. This multi-functional controller achieves the same adaptability as multiple separate controllers without the associated response delay penalty.

Inventive Principle:
Principle #6Universality (Multi-functionality)

2Reliability

If evaluation functions different from each other are used in multiple extreme value search controllers, then interference avoidance can be achieved, but the spatial margin for interference avoidance decreases

Engineering Contradiction:
Improveinterference avoidanceVSAvoidspatial margin
Core Design Contradiction:
ReliabilityVSArea of stationary object

Solution Approach 1:

The patent calculates multiple evaluation functions (J1, J2, J3) that represent different potential trajectory options before selecting the final trajectory. By evaluating multiple possibilities in advance and selecting the one that maximizes spatial margin, the system ensures both reliable interference avoidance and adequate spatial margin.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The system uses feedback from the evaluation function calculations to adjust the trajectory selection. The evaluation functions provide information about the spatial margin and interference risk, and this feedback is used to select trajectories that maintain adequate spatial margins while avoiding interference objects.

Inventive Principle:
Principle #23Feedback

Data Source

PatentUS12179755B2Trajectory generation device and trajectory generation method
Publication Date: 2024.12.31 HONDA MOTOR CO LTD
  • US12179755B2 patent drawing
  • US12179755B2 patent drawing
  • US12179755B2 patent drawing

AI summary

A trajectory generation device 1 determines first to third predicted trajectories by using model formulas (1) to (20) that model a trajectory of a host vehicle 3 on the basis of peripheral information of an information detection device 4, calculates a risk potential R_p_mi_vj_v_r by using the first to third predicted trajectories, and determines moving average values P_c1 to P_c3 so that the risk potential R_p_mi_vj_v_r decreases. In a case where the moving average values P_c1 and P_c2 have the same sign, the first predicted trajectory is determined while reflecting the moving average value P_c2 for correcting the second predicted trajectory.