Robotic Contact Control Using State Estimation for Object Grasping
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional robotic systems struggle to autonomously perform complex manipulation tasks outside controlled environments due to reliance on known object states and assumptions that are not feasible with state-of-the-art sensors, leading to ineffective interaction with various objects in dynamic environments.
Innovation Solution
A robotic control system that utilizes a database and software modules to model objects using rigid body kinematics and dynamics, processing sensory data and camera images to identify objects, estimate their states, determine grasping configurations, and communicate motor forces to the robot for accurate manipulation, enabling autonomous interaction with diverse objects.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If conventional robotic systems use traditional control algorithms with search algorithms for motion planning, then they can perform sophisticated manipulation tasks in controlled environments, but they fail to autonomously perform tasks in unstructured environments due to reliance on known object states and unrealistic assumptions
Solution Approach 1:
The system continuously updates the object state estimate by incorporating sensory data from cameras and force-torque sensors during manipulation. The state estimator uses feedback from both visual sensors and tactile force sensors to refine the object's position, orientation, and contact state in real-time, enabling autonomous adaptation to unstructured environments.
Solution Approach 2:
The system performs preliminary sensing and state estimation before manipulation tasks begin. By pre-identifying objects and estimating their initial states using sensory data, the system prepares the necessary information for subsequent autonomous manipulation without requiring predefined object knowledge.
2Reliability
If simulation-based control algorithms assume perfect knowledge of object states and ignore sensory systems, then robots can perform complex tasks in virtual environments, but this approach is not feasible with state-of-the-art sensors in real-world applications
Solution Approach 1:
The system introduces a state estimator as an intermediary component that bridges the gap between imperfect sensor measurements and the control algorithm's needs. This estimator fuses information from multiple sensors (cameras, force-torque sensors) and uses physical models to infer accurate object states, compensating for individual sensor limitations and providing reliable state information to the control system.
Solution Approach 2:
The system dynamically adjusts the parameters used in state estimation based on the quality and availability of sensory data. By adapting estimation parameters according to actual sensor performance and environmental conditions, the system maintains reliable task execution despite variations in measurement precision.
3Productivity
If robots are provided with greater accuracy of objects' states than obtainable using state-of-the-art sensors, then search algorithms can find satisfactory solutions, but this assumption cannot be met in real-world sensing conditions
Solution Approach 1:
The system replaces reliance on high-precision mechanical sensing with a computational state estimation approach. Instead of depending on sensors to directly provide accurate object states, the system uses software-based estimation that combines imperfect sensor readings with physical models to infer object states, thereby maintaining productivity without requiring unrealistic sensor precision.
4Measurement precision
If engineers control the environment to favor sensing by controlling lighting and object placement, then robots can perform well in controlled settings, but this reduces adaptability to unstructured environments
Solution Approach 1:
The system employs a universal state estimation framework that works across diverse environments without requiring environmental control. The multi-functional approach combines multiple sensing modalities (visual and tactile) with physical models to achieve accurate state estimation whether objects are well-lit or poorly-lit, properly placed or randomly positioned, thereby maintaining both measurement precision and environmental adaptability.
Data Source
AI summary
A system comprises a database; at least one hardware processor coupled with the database; and one or more software modules that, when executed by the at least one hardware processor, receive at least one of sensory data from a robot and images from a camera, identify and build models of objects in an environment, wherein the model encompasses immutable properties of identified objects including mass and geometry, and wherein the geometry is assumed not to change, estimate the state including position, orientation, and velocity, of the identified objects, determine based on the state and model, potential configurations, or pre-grasp poses, for grasping the identified objects and return multiple grasping configurations per identified object, determine an object to be picked based on a quality metric, translate the pregrasp poses into behaviors that define motor forces and torques, communicate the motor forces and torques to the robot.