Grid-Based Obstacle Self-Learning for Autonomous Lawn Mowing Robots
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for lawn mowing robots to detect obstacles are either complex and non-adaptive, requiring repetitive teaching for changes in the environment, or inefficient in real-time path planning, leading to reduced mowing efficiency and incomplete mowing.
Innovation Solution
A self-learning method that automatically acquires and stores boundary data of obstacles without external input, allowing the robot to autonomously traverse a grid map, mark traversed grids, detect collisions, and determine obstacle-free paths for efficient navigation and adaptation to new obstacles.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If obstacle information is acquired by pre-teaching, then the robot can avoid obstacles, but the teaching process is complex and cannot adapt to changing environment
Solution Approach 1:
The robot performs self-learning by autonomously traversing the grid map and identifying obstacles through collision detection, eliminating the need for external teaching processes. The system learns obstacle locations independently during normal operation.
Solution Approach 2:
The robot pre-traverses the entire grid map before actual mowing operations to build a complete obstacle database. This preliminary exploration phase prepares the navigation system with advance knowledge of obstacle locations.
2Adaptability or versatility
If obstacle information is acquired by real-time sensor detection, then the robot can adapt to new obstacles, but path planning becomes dynamic reducing mowing efficiency
Solution Approach 1:
The robot performs preliminary traversal of the grid map to detect and record all obstacles before actual mowing operations. This advance detection allows the use of static path planning during mowing, maintaining high efficiency while adapting to new obstacles.
Solution Approach 2:
The mowing process is divided into two distinct phases: a preliminary traversal phase for obstacle detection and learning, and a subsequent mowing phase for efficient operation. This segmentation allows the robot to gather environmental information without compromising mowing productivity.
3Reliability
If real-time sensor detection is used, then new obstacles can be detected, but continuous collision detection reduces mowing efficiency
Solution Approach 1:
The robot performs a preliminary traversal phase where it systematically detects and records all obstacles in the environment before beginning actual mowing. This one-time detection phase eliminates the need for continuous collision detection during productive mowing operations.
Solution Approach 2:
After the preliminary obstacle detection phase, the robot maintains continuous mowing operations using the pre-acquired obstacle information for static path planning, ensuring uninterrupted productive action without the need for continuous sensing and dynamic replanning.
Data Source
AI summary
A self-learning method for an obstacle and a self-learning method for a new obstacle are provided. With the self-learning method, in step 1, data of a grid map in a movement region is acquired, and each of grids of the grid map is marked as not being traversed. In step 2, a starting grid of the grid map is set, data of the starting grid is pushed into a stack, a movement unit is controlled to move to the starting grid, and the starting grid is remarked as being traversed. In step 3, the movement unit is controlled to traverse the grid map, and boundary data of the obstacle is acquired based on marks of the grids in the grid map. In step 4, boundary data of the obstacle is stored.