Semantic Object Mapping for Robots in Dynamic Environments
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing robotic mapping technologies fail to accurately locate specific objects within environments over time due to their static nature, lacking consideration for object characteristics and transient locations, which leads to inefficient navigation and object retrieval.
Innovation Solution
The implementation of lifelong semantic mapping, which updates object probabilities based on temporal and spatial relationships, object classes, and user corrections, allowing robots to navigate more effectively by identifying the most likely locations of objects based on their characteristics and usage patterns.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If static mapping is used to represent environment, then mapping simplicity is maintained, but object location accuracy deteriorates over time
Solution Approach 1:
The patent transforms static environment mappings into dynamic mappings that continuously update object location probabilities over time. The system maintains probability distributions for object locations and updates them based on temporal characteristics, making the mapping adaptive to object movement while preserving computational efficiency through probabilistic models.
Solution Approach 2:
The patent changes the parameter representation from fixed coordinates to time-varying probability distributions. By representing object locations as probability functions that evolve over time based on object characteristics and observation data, the system improves location accuracy while maintaining manageable complexity through parametric models.
2Use of energy by moving object
If traditional mapping methods are used, then computational resources are conserved, but object retrieval efficiency deteriorates
Solution Approach 1:
The patent performs preliminary actions by pre-computing and maintaining probability distributions for object locations based on their temporal characteristics. This allows the robotic system to quickly query likely object locations without performing exhaustive searches, improving retrieval efficiency while keeping computational costs manageable through advance probability modeling.
Solution Approach 2:
The patent replaces traditional mechanical search methods with information-based probabilistic reasoning. Instead of physically searching through the environment to locate objects, the system uses computed probability distributions to directly identify likely object locations, substituting computational inference for physical exploration and improving retrieval efficiency.
3Device complexity
If static probability values are used for object locations, then mapping computation is simplified, but adaptability to object movement deteriorates
Solution Approach 1:
The patent implements dynamic probability values that automatically adapt to object movement based on temporal characteristics. The system maintains probability functions for each object location that evolve over time according to object-specific parameters, enabling automatic adaptation to movement patterns without complex real-time tracking computations.
Data Source
Figure 1A
Figure 1B
Figure 2
AI summary
Methods, systems, and apparatus for receiving a reference to an object located in an environment of a robot, accessing mapping data that indicates, for each of a plurality of object instances, respective probabilities of the object instance being located at one or more locations in the environment, wherein the respective probabilities are based at least on an amount of time that has passed since a prior observation of the object instance was made, identifying one or more particular object instances that correspond to the referenced object, determining, based at least on the mapping data, the respective probabilities of the one or more particular object instances being located at the one or more locations in the environment, selecting, based at least on the respective probabilities, a particular location in the environment where the referenced object is most likely located, and directing the robot to navigate to the particular location.