AI-based robotic grasp
Robotic grasp using neural networks and several stochastic machine learning
- Research Background
- Research Objectives
- Research Output
1. Object detection
Recognizing objects using Mask R-CNN on given RGB camera images.
* Mask R-CNN
1) Bounding box detection based on Faster R-CNN and segmented objects
2) Using ResNet and RPN (Region Proposal Network)
3) ResNet to extract the feature map of given RGB camera images.
4) RPN (Region Proposal Network) to find the bounding boxes of objects
< Example of Mask R-CNN >
< Structure of Mask R-CNN >
* Fine-tuned Mask R-CNN
1) 45 kinds of objects
< Examples of target objects: ACRV picking benchmark dataset >
2) Demonstration < Fine-tuned Mask R-CNN>
2. Object Grasping based on ANN
* Learning grasping pose of robot
* Disadvantages collecting training data in a real working environment
1) A lot of time
2) Robot operation cost
3) Supervision
* Implementing a realistic working environment on the simulator
1) Collecting large amounts of data through simulation
< System configuration: (a) robot in simulation and (b) real robot >
* Learning to grasp based on AlexNet
1) Input: image of the detected object
2) Output: pose of gripper with the highest grasping success rate
< Structure of fine-tuned AlexNet >
* The ensemble learning method combines various classifiers to achieve better performance than a single classifier.
1) Combining the results of learning models
2) Obtaining more reliable results than a learning model
< Concept of ensemble learning >
* 4 classifiers based on four data sets with ensemble learning: trained under
1) Real data
2) Simulation data
3) Another simulation data
4) Real data + simulation data
< Application of ensemble learning to grasp >
< Example of object grasping >
3. Object Grasping based on stochastic machine learning
* Object pose (x, y, q) is necessary to grasp objects.
* Location (x, y) can be known using Mask R-CNN
* Angle (q) is predicted by using the masks of Mask R-CNN and PCA (principal component analysis)
* Extracting depth images using masks
< Example of extracted depth image using mask >
* Grasping pose estimation based on PCA
1) Predicting the shortest direction (minor axis) of the objects at the center of depth images
< Result of predicted grasping pose >
* Grasping demonstration
< Example of object grasping >
4. Bin picking demonstration for 3 scenarios * Target objects: 20 known objects + 5 unknown objects * Known objects: 20 objects = 7 foods + 7 toys + 6 tools - Stage 1: Objects in blue boxes - Stage 2: All objects - Stage 3: Objects in red boxes + 5 unknown objects
* Scenarios
* Demonstration for each stage
AI-based robotic assembly
- Research Background
- Research Objective
- Research Outputs
1) Reinforcement learning method based on DMP and PoWER
2) Reinforcement learning method based on NNMP and DDPG
- Reinforcement learning method based on DMP and PoWER
< Control system using DMP & PoWER >
* Dynamic movement primitive (DMP)
1) A motor primitive based on Stefan Schaal's proposed dynamical system
2) DMP can generate complex trajectories using the minimum number of linear parameters.
3) The shape of the trajectory is determined by the linear parameter. àSuitable for reinforcement learning
* Policy learning by weighting exploration with the returns (PoWER)
1) Episode based reinforcement learning algorithm applicable to linear deterministic policy function
2) Reinforcement learning algorithm using expectation maximization (EM) à No learning rate required.
3) PoWER generally has an excellent learning speed, but the applicable form of the policy function is limited
* Demonstration
1) Robot: SCORA-V (Safe Collaborative Robot Arm – Vertical type) developed in the laboratory
2) Control system: PC-based controller with a current control cycle of 1 ms through the EtherCAT communication
3) Force control algorithm: torque-based impedance controller
4) Assembly parts: square peg-in-hole, size: 50.0 x 50.0 x 30.0 mm, tolerance: 0.1 mm
< Assembly demo before and after learning >
- Reinforcement learning method based on NNMP and DDPG
< Control system using NNMP & DDPG >
* Neural network-based movement primitive (NNMP)
1) DNN is used to generate complex trajectories by using various input signals (measured force and position).
2) The velocity and position are calculated by integrating the acceleration to generate a continuous trajectory.
3) The size and motion time of the trajectory can be changed by adjusting the normalization matrix.
4) DAgger based Imitation learning algorithm for proposed NNMP is developed.
< Neural network-based movement primitive >
* Deep deterministic policy gradient (DDPG) for NNMP
1) The measured force and position are added as the state of the NNMP to reflect the contact state.
2) In order to apply DDPG, the neural network of NNMP is regarded as an actor network.
3) Ornstein–Uhlenbeck (OU) noise is added to the action for exploration in reinforcement learning.
< Structure of robot system for reinforcement learning with force controller and NNMP >
* Demonstration
1) Robot: SCORA-V (Safe Collaborative Robot Arm – Vertical type) developed in the laboratory
2) Control system: PC-based controller with a current control cycle of 1 ms through the EtherCAT communication
3) Force control algorithm: torque-based impedance controller
4) Assembly parts: square peg-in-hole, size: 50.0 x 50.0 x 30.0 mm, tolerance: 0.1
mm < Assembly demo before and
after learning >
Simulator to the real-world transfer of manipulation policy - Research Background
- Research Objective
- Research Outputs
* A demonstration data collection system for aggregating robot’s motion Markov transitions < Transition data collection via velocity kinematics >
* Asymmetric Actor-Critic based for manipulating real-world robot in POMDP
|