Self-Location Re-Estimation Control After Unpredictable Position Change
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing technologies face challenges in accurately estimating a mobile object's self-location when it is unknown, particularly after external forces cause a disruption in continuous location information, leading to loss of recognition and inaccurate detection.
Innovation Solution
A control device with a self-position detection unit, a position change detection unit, and a self-location estimation unit that employs multiple estimation models, including a time-series information accumulation unit and a Kalman filter, to estimate the self-location using both past and current sensor information, switching between models based on detected position changes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a single estimation model is used to estimate self-location continuously, then the estimation process is simple, but the accuracy deteriorates when position changes unpredictably occur
Solution Approach 1:
The system dynamically switches between a first estimation model and a second estimation model based on whether unpredictable position changes are detected. When such changes occur, the system transitions from continuous estimation using one model to using an alternative model, allowing the estimation approach to adapt to changing conditions and maintain accuracy
Solution Approach 2:
The system changes the estimation parameters by switching between different estimation models. The first estimation model is used under normal conditions, while the second estimation model is activated when unpredictable position changes are detected, effectively changing the estimation approach to match the current state
2Reliability
If correction after kidnap reflects movement distance during kidnap, then the self-location can be corrected, but the self-location cannot be detected accurately due to loss of continuity
Solution Approach 1:
The estimation process is segmented into two distinct models: a first estimation model for normal continuous estimation and a second estimation model for when unpredictable position changes occur. This segmentation allows each model to be optimized for its specific condition, improving overall reliability without sacrificing accuracy
3Ease of operation
If monitoring is limited to fluctuation between control input and actual movement, then the system is simple to operate, but the self-location cannot be estimated again using multiple sensors
Solution Approach 1:
The system is designed to handle multiple estimation scenarios using a universal framework that can process both continuous position changes and unpredictable position changes. The estimation unit can operate in multiple modes, making it versatile enough to handle various sensor inputs and estimation requirements without requiring separate monitoring systems
Data Source
AI summary
A control device and a control method can quickly estimate a self-location even when the self-location is unknown. In a case of storing information supplied in a time series detected by LIDAR or a wheel encoder and estimating a self-location by using the stored time-series information, when a position change happens unpredictably in advance such as a kidnap state is detected, the stored time-series information is reset, and then the self-location is estimated again. Example host platforms include a multi-legged robot, a flying object, and an in-vehicle system that autonomously moves in accordance with a mounted computing machine.