Task Planner for Cluttered Object Retrieval
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Robotic manipulation tasks in cluttered environments are complicated by dense clutter, making it difficult for robots to identify and access target objects, and existing user interfaces for assistive devices are often not intuitive, especially for individuals with disabilities.
Innovation Solution
A system that includes an image capture device and a processor-based task planner using an action prediction model and dynamic prediction model to determine the most probable actions for object retrieval, simulating subsequent states, and implementing these actions through a motion planner and actuator, which can perform operations like grasping or placing objects.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If traditional joysticks or simple control interfaces are used for assistive devices, then device control is achieved, but the interface is not intuitive and difficult for users with disabilities to operate
Solution Approach 1:
The system performs autonomous task planning and motion planning without requiring complex user inputs. The robot independently calculates probabilities of success for potential actions, selects optimal actions, and executes multi-step tasks to retrieve objects, making the system self-sufficient and reducing the burden on users with disabilities
Solution Approach 2:
The patent replaces traditional mechanical control interfaces (joysticks) with an intelligent software-based system that uses action prediction models, dynamic prediction models, and probabilistic reasoning to automatically plan and execute tasks, substituting mechanical interaction with cognitive automation
2Adaptability or versatility
If robots operate in cluttered environments, then they can access more real-world scenarios, but dense clutter obscures target objects making retrieval difficult
Solution Approach 1:
The system performs preliminary task planning by determining multiple potential actions and calculating their probabilities of success before execution. The motion planner pre-computes feasible motion sequences considering clutter obstacles, and the system simulates subsequent states to predict outcomes, preparing in advance for complex retrieval scenarios in cluttered environments
Solution Approach 2:
The patent introduces an action prediction model and dynamic prediction model as intermediaries between the robot's perception system and execution system. These models process visual information about cluttered environments, predict the effects of potential actions, and guide the motion planner to successfully retrieve obscured target objects
3Adaptability or versatility
If complex multi-step tasks are required for object retrieval, then task capability is improved, but it becomes difficult for users to execute and control
Solution Approach 1:
The patent segments complex retrieval tasks into discrete potential actions with calculable probabilities of success. Each action is evaluated independently by the action prediction model, and the system selects the optimal sequence through probabilistic reasoning, breaking down complex tasks into manageable, evaluable units that can be autonomously planned and executed
Data Source
AI summary
Systems and techniques for target object retrieval may include or utilize an image capture device, and a task planner. The image capture device may receive an image of an environment including identified objects. The task planner may determine potential actions, calculate a probability of success of achieving a desired goal for each of the potential actions based on an action prediction model, the corresponding potential action, a current state of the environment, any previously taken action, and the desired goal, select a potential action associated with the highest calculated probability of success, and simulate a subsequent state based on the selected potential action and a dynamic prediction model. The potential actions may be associated with an identified object of the identified objects and an operation to be performed on the identified object.