Adaptive Robot Stacking Planning With Learned Preconditions

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Conventional task and motion planning (TAMP) systems for object-invariant stacking operations are inefficient and require manual encoding of object-specific preconditions, leading to human errors and challenges in performing diverse stacking tasks.

Innovation Solution

An adaptive task and motion planning (ATAMP) system that learns and adapts to new preconditions using a virtual Discrete Action Space (DAS) and an n-armed bandit problem, allowing robotic agents to perform object-invariant stacking operations by generating and updating preconditions based on real-time interactions.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If manual encoding of object-specific preconditions is used in conventional TAMP systems, then the system can perform stacking operations, but the system requires tremendous and tedious effort including human errors and cannot handle diverse stacking tasks

Engineering Contradiction:

Improveability to handle diverse stacking tasksVSAvoidmanual encoding effort and human errors

Core Design Contradiction:

Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The system performs self-learning by automatically acquiring new preconditions through interaction with the environment. The robotic agent independently identifies and adds new preconditions to the action model without human intervention, enabling the system to adapt to diverse stacking tasks while eliminating the need for manual encoding effort

Inventive Principle:

Principle #25Self-service

Solution Approach 2:

The system dynamically changes the preconditions parameters in the action model based on learned experiences. By modifying the precondition set through learning new object-specific requirements, the system adapts its behavior to handle various stacking tasks without requiring manual reconfiguration

Inventive Principle:

Principle #35Parameter changes

2Reliability

If conventional TAMP systems use fixed preconditions for stacking operations, then the system structure is simple, but the system fails to perform satisfactorily on object-invariant stacking tasks with different objects

Engineering Contradiction:

Improvesuccess rate of stacking operationsVSAvoidgeneralization to various objects

Core Design Contradiction:

ReliabilityVSAdaptability or versatility

Solution Approach 1:

The system uses feedback from stacking operation outcomes to learn and update preconditions. When the robotic agent attempts stacking operations and observes results, it uses this feedback to identify new preconditions that improve success rates across different objects, enabling reliable performance through continuous adaptation

Inventive Principle:

Principle #23Feedback

Solution Approach 2:

The system performs preliminary learning of preconditions before executing stacking tasks. By pre-acquiring knowledge about object-specific requirements through interaction, the system prepares the action model in advance, ensuring reliable stacking operations when actual tasks are performed

Inventive Principle:

Principle #10Preliminary action

3Manufacturing precision

If the action model is enhanced with more preconditions to improve stacking success, then the task planning becomes more accurate, but the manual encoding process becomes more tedious and error-prone

Engineering Contradiction:

Improveaccuracy of task planningVSAvoidcomplexity of precondition encoding

Core Design Contradiction:

Manufacturing precisionVSDevice complexity

Solution Approach 1:

The system automatically acquires and encodes preconditions through self-learning from environmental interactions. The robotic agent independently identifies necessary preconditions for successful stacking and adds them to the action model, achieving accurate task planning without manual encoding effort or human errors

Inventive Principle:

Principle #25Self-service

Solution Approach 2:

The system replaces manual encoding processes with automated learning mechanisms. Instead of human operators manually specifying preconditions, the system uses computational learning to automatically acquire and integrate precondition knowledge, substituting mechanical human effort with automated intelligence

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Data Source

PatentUS20250289137A1Method and system for adaptive task and motion planning (ATAMP) for performing object-invariant stacking operations

Publication Date: 2025.09.18 TATA CONSULTANCY SERVICES LTD

AI summary

This disclosure relates generally to a method and system for adaptive task and motion planning (ATAMP) for object-invariant stacking operations. Traditional TAMP considering only the preconditions for execution of an object stacking task is challenging for performing all kinds of the object-invariant stacking operations The disclosed method adopts an action model for a new object and stacking is performed by drawing inferences and learning rewards using a virtual Discrete Action Space (DAS) based on a heuristically defined reward function. These inferences are utilized for identifying a plurality of new preconditions. Additionally, an efficient stacking position selection strategy is used for a n-armed bandit problem, which leads to fast convergence for performing the object-invariant stacking operations. A robotic agent repetitively interacts with an environment in real-time to adapt the action model for the new object. After adaptation, the robotic agent can perform the object-invariant stacking operations on objects with varying poses.