Description

The Caltech Robot Manipulation club aims to research and develop methods for robotic grasping and manipulation with the intent to participate and compete in a variety of competitive events.

We hope to expose undergraduate students to hands-on work in this niche area of robotics while also presenting opportunities for research, mentorship, and collaboration between undergraduates and graduates, alike.

We plan on taking a 'learning' based approach to robotics by learning and implementing policies in machine learning to perform various robotics tasks such as bin-picking, dexterous manipulation, cloth manipulation, and more! Specifically, we are interested in competing in grasping & manipulation competitions such as https://cse.usf.edu/~yusun/rgmc/2024.html and https://arm.robocup.org/. We also plan on hosting seminar series and presentations with club members and guest speakers from different universities working on reinforcement learning, control theory, 3D representations, sim2real methods, perception or robotics in general.

Table of contents

Interest

We sent out a brief interest form to Caltech Graduate and Undergraduate Students and so far we have gained 75 responses with 67 being undergraduate and 5 being graduate students. We hope to foster collaboration between undergraduate and graudate students working on the niche field of robotic manipulation. We hope to work with the ARM lab at Caltech, which is overseen by Dr. Burdick who’s also an advisor to this club, with access to the Lead Hand and the Franka Research Arm 3 with Allegro Hand mounted.

Competitions

One of the main focus of the club is getting students to work on robotic manipulation competitions such as:

10th Robotic Grasping and Manipulation Competition:

While this hasn’t been formally announced yet, based on the historical trends we expect this to come out soon with a deadline in Late January/Early February

A brief description about the competition:

The competition comprised of two main tracks in the previous season:

• Essential Track

• Manufacturing Track

We are planning to focus on tasks with lower barrier to hardware access before we get into more involved competitions.

One of the competitions that has a lower barrier is the Cloth Manipulation subtrack within the Essential Track. The objective of the competition is to localise good grasp poses in colour and depth images of garments held in the air by one gripper and is mainly a perception and algorithm problem.

The second subcompetition is In-hand Manipulation challenge which involves in-hand manipulation skills of different robot hands and various manipulation planning and control algorithms and significiant sim2real challenges. We aim to develop our policies in simulation and test it on the Leap Hand.

The third subcompetition is Picking in clutter which is the classical bin-picking challenge and involves planning and perception. We aim to develop our policies in simulation as well and test it on Franka Research Arm with Allegro Hand

Seminar Series

To make this club engaging, we also aim to invite guest speakers including PhD students and postdocs working in this domain to present their findings and invite undergraduate and graduate students to have a conversation about their research.

Involvement

We aim to have subteams working on different components of the competion such as Algorithms, Perception, Simulation, Hardware, Fundraising

People

We are led by Ryan Rudes (Co-President, rrudes@caltech.edu), Dhruv Sheth (Co-President, dsheth@caltech.edu), and Frank Xiao (Treasurer, fyxiao@caltech.edu)

OurW faculty advisor is Dr. Burdick.