Mobile Robot Path Planning for Moving Obstacle Avoidance

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current mobile robots face inefficiencies in mobility performance due to the need for lengthy calculations to generate both moving and avoidance paths, as existing map information only accounts for stationary obstacles, neglecting moving obstacles like people or other mobile robots.

Innovation Solution

A neural network machine learning method that allows for the simulation of stationary and moving obstacles in a virtual space, enabling the accumulation of teaching data for generating smooth autonomous movement paths, where users can designate paths to avoid both types of obstacles, thereby reducing the time required for path calculation.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If analytical methods are used to generate both moving path and avoidance path, then path planning accuracy is improved, but calculation time increases significantly

Engineering Contradiction:
Improvepath planning accuracyVSAvoidcalculation time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The path planning problem is segmented into two distinct components: moving path generation (based on map information) and avoidance path generation (reactive response to moving obstacles). This segmentation allows each component to use optimized algorithms appropriate to its specific requirements, reducing overall calculation time while maintaining accuracy.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The moving path is generated in advance using analytical methods based on map information before the robot begins movement. This preliminary action allows complex path planning to be performed offline, so that during actual operation, the robot only needs to perform lighter-weight avoidance path calculations when detecting moving obstacles.

Inventive Principle:
Principle #10Preliminary action

2Device complexity

If map information only includes stationary obstacles, then map data simplicity is improved, but robot mobility performance deteriorates due to inability to handle moving obstacles

Engineering Contradiction:
Improvemap data complexityVSAvoidmobility performance
Core Design Contradiction:
Device complexityVSProductivity

Solution Approach 1:

The system transitions from static map information to dynamic obstacle detection and response. While map data remains simple (containing only stationary obstacles), the robot employs real-time detection mechanisms that identify moving obstacles and trigger reactive avoidance path generation, enabling the system to handle dynamic environments without complicating the base map data structure.

Inventive Principle:
Principle #15Dynamics

Solution Approach 2:

The robot autonomously detects moving obstacles and generates avoidance paths without external intervention. The system self-manages the transition from simple map-based navigation to complex avoidance maneuvers, maintaining mobility performance while keeping map data simple.

Inventive Principle:
Principle #25Self-service

3Measurement precision

If analytical generation of avoidance path is performed every time moving obstacle is detected, then obstacle avoidance accuracy is improved, but real-time response capability deteriorates

Engineering Contradiction:
Improveavoidance path accuracyVSAvoidresponse speed
Core Design Contradiction:
Measurement precisionVSSpeed

Solution Approach 1:

The system changes the parameters used for path generation based on the situation. For the moving path, analytical methods with high accuracy parameters are used. For avoidance paths triggered by moving obstacles, the system uses optimized parameters that balance accuracy with computational speed, allowing real-time response while maintaining adequate avoidance precision.

Inventive Principle:
Principle #35Parameter changes

Data Source

PatentEP3757714B1Machine learning method and mobile robot
Publication Date: 2022.10.26 TOYOTA JIDOSHA KK
  • EP3757714B1 patent drawingFigure 1
  • EP3757714B1 patent drawingFigure 2
  • EP3757714B1 patent drawingFigure 3

AI summary

A neural network machine learning method include: a first arrangement step of arranging a stationary first obstacle and a moving second obstacle in a virtual space; a second arrangement step of arranging a current position and a destination of a mobile robot (100) in the virtual space; a movement step of making the second obstacle move in accordance with a predetermined condition; and a reception step of receiving, from a user, a designation of a moving path for the mobile robot (100) to go to the destination from the current position while avoiding the stationary first obstacle and the moving second obstacle, in which learning is performed using teaching data accumulated by repeatedly executing the above steps.