Action Recommendation for Shared Learning in Autonomous Mobile Bodies

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Leg type mobile robots struggle to efficiently share learning results with other autonomous mobile bodies, limiting their ability to improve learning processes.

Innovation Solution

An information processing apparatus and method that includes an action recommendation unit, which presents recommended actions to autonomous mobile bodies based on action histories from multiple robots and situation summaries, enabling more effective action planning.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If autonomous mobile bodies perform independent learning individually, then each body can independently improve its actions, but the learning results cannot be shared with other autonomous mobile bodies

Engineering Contradiction:

Improvelearning capabilityVSAvoidlearning result sharing

Core Design Contradiction:

Adaptability or versatilityVSLoss of information

Solution Approach 1:

The patent merges the learning systems of multiple autonomous mobile bodies by collecting action histories from multiple bodies and using a centralized learning model to generate shared situation summaries and recommended actions. This allows learning results to be combined and distributed across the network, resolving the contradiction between independent learning and result sharing.

Inventive Principle:

Principle #5Merging (Combining)

Solution Approach 2:

The patent introduces an intermediary system that collects action histories from multiple autonomous mobile bodies, processes this data through learning models, and distributes situation summaries and recommended actions back to the bodies. This intermediary mechanism enables information sharing while preserving the independence of each mobile body's learning process.

Inventive Principle:

Principle #24Intermediary (Mediator)

2Ease of operation

If action plans are based solely on individual situation estimation, then each autonomous mobile body can act independently, but the quality of action recommendations is limited by individual experience

Engineering Contradiction:

Improveindependent action capabilityVSAvoidaction recommendation quality

Core Design Contradiction:

Ease of operationVSReliability

Solution Approach 1:

The patent creates a universal situation summary that aggregates experience from multiple autonomous mobile bodies, making the learning model applicable to various situations encountered by different bodies. This universal knowledge base improves recommendation quality while maintaining individual independence through the situation estimation framework.

Inventive Principle:

Principle #6Universality (Multi-functionality)

Solution Approach 2:

The patent implements a feedback loop where action histories from multiple bodies are collected, processed into situation summaries, and used to generate recommended actions that are fed back to the autonomous mobile bodies. This feedback mechanism continuously improves action recommendation quality based on collective experience while preserving independent operation.

Inventive Principle:

Principle #23Feedback

Data Source

PatentUS11675360B2Information processing apparatus, information processing method, and program

Publication Date: 2023.06.13 SONY GROUP CORP

AI summary

There is provided an information processing apparatus and an information processing method that can provide more useful information for an action plan of an autonomous mobile body, the information processing apparatus including an action recommendation unit configured to present a recommended action recommended to an autonomous mobile body, to the autonomous mobile body that performs an action plan based on situation estimation. The action recommendation unit determines the recommended action on the basis of an action history collected from a plurality of the autonomous mobile bodies, and on the basis of a situation summary received from a target autonomous mobile body that is a target of recommendation. The information processing method includes presenting, by a processor, a recommended action recommended to an autonomous mobile body, to the autonomous mobile body that performs an action plan based on situation estimation.