Tactile Object Recognition Using Interactive Grasp Feedback
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Robotic manipulation systems face challenges in classifying novel object instances without prior knowledge of object classes or labels, as existing tactile recognition methods require pre-training and rely heavily on vision-based systems that are blind during physical contact and difficult to estimate object states post-grasping.
Innovation Solution
A novel approach using 3D tactile descriptors and One-Class SVM for unsupervised learning, enabling classification of novel objects through interactive tactile feedback without pre-training or ground truth labels, leveraging low-resolution tactile sensor arrays and high-resolution camera-based sensors.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If supervised learning with training datasets is used for object classification, then classification accuracy is improved, but the system cannot recognize novel object instances without pre-training
Solution Approach 1:
The system performs preliminary tactile exploration actions (probing, grasping, manipulating) to collect data about novel objects, enabling the robot to build a model of the object's properties through active interaction rather than passive observation
Solution Approach 2:
The system uses tactile feedback from force sensors and moment sensors during manipulation to continuously update its understanding of object properties, allowing adaptive recognition of novel objects based on real-time interaction data
2Measurement precision
If machine vision systems are used for object recognition, then object localization and identification are improved, but the system remains blind during physical contact and cannot estimate object state post-grasping
Solution Approach 1:
The system merges vision-based pre-grasp localization with tactile feedback during grasping and manipulation, combining the strengths of both modalities to maintain continuous awareness of object state throughout the manipulation task
Solution Approach 2:
Tactile sensors serve as an intermediary between the robot's grasping actions and object state estimation, providing direct physical contact information that bridges the gap left by vision systems during closed-loop manipulation
3Measurement precision
If high-resolution tactile sensors are used for detailed surface detection, then surface feature recognition is improved, but device complexity and cost increase
Solution Approach 1:
The system applies partial tactile contact through selective probing and grasping actions, using force sensors and moment sensors to detect sufficient surface features without requiring complete surface coverage by high-resolution tactile arrays
Solution Approach 2:
The system replaces complex high-resolution tactile sensor arrays with a combination of lower-resolution tactile sensors supplemented by force/moment sensors that infer surface properties through mechanical interaction and physics-based reasoning
Data Source
AI summary
A controller is provided for interactive classification and recognition of an object in a scene using tactile feedback. The controller includes an interface configured to transmit and receive the control, sensor signals from a robot arm, gripper signals from a gripper attached to the robot arm, tactile signals from sensors attached to the gripper and at least one vision sensor, a memory module to store robot control programs, and a classifier and recognition model, and a processor to generate control signals based on the control program and a grasp pose on the object, configured to control the robot arm to grasp the object with the gripper. Further, the processor is configured to compute a tactile feature representation from the tactile sensor signals and to repeat gripping the object and computing a tactile feature representation with the set of grasp poses, after which the processor, processes the ensemble of tactile features to learn a model which is utilized to classify or recognize the object as known or unknown.