Vehicle Localization Mapping with Credibilist Occupancy Grids
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current localization and mapping techniques for mobile machines in dynamic environments, such as SLAM, primarily focus on probabilistic approaches that fail to effectively account for uncertainty and absence of information, leading to inaccuracies in obstacle detection and mapping.
Innovation Solution
A device and method that utilize a credibilist approach, incorporating likelihoods of occupation, non-occupation, and non-knowledge values, with a normalization process to determine the best correspondence between sensor measurements and constructed cartography, incorporating a conflict factor to handle contradictory information.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If probabilistic approaches are used to determine occupancy probability in SLAM, then localization and mapping can be performed, but accuracy deteriorates due to failure to account for uncertainty and absence of information
Solution Approach 1:
The patent transforms the traditional probabilistic parameter (single occupancy probability) into a credibilist parameter system comprising three distinct values: occupancy likelihood, non-occupancy likelihood, and non-knowledge likelihood. This parameter transformation enables the system to represent uncertainty and absence of information explicitly, thereby improving both localization accuracy and obstacle detection reliability simultaneously
Solution Approach 2:
The patent adds a new dimension to the representation of environmental knowledge by introducing the third credibilist value (non-knowledge likelihood) alongside the traditional occupancy and non-occupancy probabilities. This dimensional expansion allows the system to distinguish between 'known empty' and 'unknown' areas, resolving the contradiction between localization precision and detection reliability
2Reliability
If traditional SLAM methods are used, then mapping can be performed, but accuracy deteriorates in dynamic environments due to inability to handle contradictory information
Solution Approach 1:
The patent introduces the non-knowledge likelihood value as an intermediary state that mediates between conflicting occupancy and non-occupancy information in dynamic environments. When contradictory sensor data is received, the system can transition to the non-knowledge state rather than making erroneous deterministic decisions, thereby maintaining mapping accuracy while adapting to environmental changes
Solution Approach 2:
The credibilist approach implements dynamic adaptability by allowing the three likelihood values to evolve continuously as new sensor data arrives. The system can dynamically adjust between occupancy, non-occupancy, and non-knowledge states based on current environmental conditions, enabling reliable mapping performance in both static and dynamic scenarios
Data Source
Figure 1~3
AI summary
The invention relates to a location and mapping device intended for being installed in a mobile vehicle in an environment comprising a plurality of zones, with: a receiving module (11) designed to receive data of the position (d(a)) of obstacles relative to the vehicle; a processing module (12, 14) designed to determine, in accordance with the data received (d(a)) and for a plurality of zones positioned relative to the vehicle, measurement values (M_SCANt,i) that respectively represent a likelihood of occupation of the zone in question, a likelihood of non-occupation of the zone in question and a likelihood of not knowing the zone in question; a mapping module (18, 20) designed to build, for each zone of said plurality of zones of the environment, values (M_GRIt,k) which respectively represent a likelihood of occupation of the zone in question, a likelihood of non-occupation of the zone in question and a likelihood of not knowing the zone in question; and a location module (16) designed to determine the position of the vehicle in the environment, which maximises a measurement of correspondence between the measurement values (M_SCANt,i) and the built values (M_GRIt,k). The invention also describes an associated method.