Robotic Dexterity
This course covers fundamental principles of robotic dexterous manipulation, including kinematics, manipulator and gripper control, motion planning, contact modeling, and grasp stability. It explores robotic perception (vision and tactile) and advanced machine learning techniques (reinforcement learning and learning from demonstration). Students will apply these concepts in a hands-on group project using robotic manipulators for complex tasks. Homework assignments with both analytical and coding components will reinforce the concepts learned in class. Students should feel comfortable in Python and Linux (dual booting on a personal computer may be necessary for some assignments).
Link to Stanford Explore Courses
Course Components
Lecture
~50min lectures, covering concepts for robotic manipulation.
Homework
Written and Coding: (Jupyter Notebook - Python).
Analytical and coding problems to reinforce concepts in robotic modeling, control, and machine learning for manipulation.
Literature Group Discussion
Group discussions, based on relevant reading.
Small group discussion, summarizing key literature.
Project
Group Project: Groups of 5-7 students work together to program a physical robotic manipulator to perform dexterous manipulation tasks.
Individual Project: Each student proposes a research project and provides an oral presentation and written report (IEEE conference format), leveraging class concepts.
Spring 2025 Group Project Description
This year, the goal of the project will be to make a robotic manipulator capable of performing a range of dexterous tasks. Our class will use the UFactory XArm 7 degree of freedom manipulator with wrist wrench sensing, arm-mounted camera and choice of gripper.
In this project, teams will perform dexterous manipulation tasks as outlined in Figure 1. The tasks are as follows:
- Pick and Place: This task involves a) visual perception: detecting the block and target final location b) Control and Motion Planning: Using motion planner in ROS to plan a safe, collision free trajectory to pick and place block. Points are awarded if the block is within the specified target region
- Peg in Hole: This task involves a) visual perception: detection of peg and hole location b) Force/Tactile Perception and Control and Motion Planning: using force feedback to determine if peg has been placed in the hole (10 min time limit).
- Coin Pickup: This task involves a) visual perception: detecting the penny on the table b) Control and Motion Planning: retrieving and placing the coin in the designated zone.
- Flat Dollar Bill Pickup: This task involves a) visual perception: detecting the dollar bill on the table b) Control and Motion Planning: retrieving and placing the bill in the designated zone.
