Course syllabus for Athletic intelligence in robotics

Course syllabus adopted 2024-12-19 by Head of Programme (or corresponding).

Overview

  • Swedish nameAtletisk intelligens i robotik
  • CodeTRA455
  • Credits7.5 Credits
  • OwnerTRACKS
  • Education cycleSecond-cycle
  • DepartmentTRACKS
  • GradingTH - Pass with distinction (5), Pass with credit (4), Pass (3), Fail

Course round 1

  • Teaching language English
  • Application code 97182
  • Minimum participants8
  • Open for exchange studentsYes

Credit distribution

0124 Project 7.5 c
Grading: TH		 7.5 c

In programmes

Examiner

Eligibility

General entry requirements for Master's level (second cycle)
Applicants enrolled in a programme at Chalmers where the course is included in the study programme are exempted from fulfilling the requirements above.

Specific entry requirements

English 6 (or by other approved means with the equivalent proficiency level)
Applicants enrolled in a programme at Chalmers where the course is included in the study programme are exempted from fulfilling the requirements above.

Course specific prerequisites

In addition to the general requirements to study at the second-cycle level at Chalmers, necessary subject or project specific prerequisite competences (if any) must be fulfilled. Alternatively, the student must obtain the necessary competences during the course. The examiner will formulate and check these prerequisite competences. The student will only be admitted in agreement with the examiner.

It is preferable, but not essential, to have a good understanding of mechanics, control theory, linear algebra, advanced calculus, numerical methods, mechatronics, programming etc.

Aim

The course provides a platform to work and solve challenging cross-disciplinary authentic problems from different stakeholders in society such as the academy, industry or public institutions. Additionally, the aim is that students from different educational programs practice working efficiently in multidisciplinary development teams

This course aims to provide the fundamentals of athletic intelligence in robotics to enable robust sensorimotor control in robots. Students will learn to put this knowledge into practice during tutorials and in exercise sheets using Python implementation and robot simulations and/or hardware experiments.

Learning outcomes (after completion of the course the student should be able to)

  • critically and creatively identify and/or formulate advanced architectural or engineering problems
  • master problems with open solutions spaces which includes to be able to handle uncertainties and limited information.
  • lead and participate in the development of new products, processes and systems using a holistic approach by following a design process and/or a systematic development process.
  • work in multidisciplinary teams and collaborate in teams with different compositions
  • orally and in writing explain and discuss information, problems, methods, design/development processes and solutions
  • Explain the concept of athletic intelligence and identify its key aspects.
  • Model an athletic robot as a dynamical system and apply optimization and/or learning-based tools to generate complex behavior.
  • Develop model-based controllers which can stabilize the behavior of an athletic robot in the presence of disturbances.
  • Perform stability analysis of controllers.
  • Apply methods for system identification and perform state estimation
  • Identify open challenges in robotics research and current trends in state-of-the-art.
  • Communicate confidently using the terminology in the field of robotics.

Content

Traditional robots today (such as the ones used in factories) have a fixed base and are fully actuated under their operating conditions. However, modern robots inspired by animals (such as hoppers, quadruped, humanoids) are not bound to one place and are always under-actuated. Like animals, these robots can perform dynamic movements, demonstrate compliance, and are robust to contact during their movements. Robots of the future will be able to move more dynamically and safely in a rugged environment shared with humans. To develop such robots, it is crucial to focus on the athletic intelligence in robots. The course includes practical examples demonstrating how the theory applies to modern robotics. By the end of the course, students will gain valuable insights into behavior generation and control strategies used in contemporary athletic robots, such as Atlas (Boston Dynamics), Digit (Agility Robotics), and the Unitree Go2 quadruped.

Organisation

The course is run by a teaching team responsible for lectures, tutorials and consultancy sessions. The course is organized into five lecture blocks. At the end of each block, the students will be asked to submit a group assignment which tests their knowledge and skills gained from the lecture block. At the end of the course, individual assessment of students will be made by oral examination.

The course on Athletic Intelligence in Robotics is structured to provide a comprehensive understanding of the field through a blend of theoretical lectures, practical tutorials, consultancy sessions, and group assignments. The course is divided into five blocks, each focusing on a specific aspect of athletic intelligence in robotics.
In Block 1: Introduction to Athletic Intelligence, students are introduced to the fundamental concepts of athletic intelligence and building blocks of contemporary athletic robots. The lectures cover topics such as underactuated systems, sensorimotor control, and the physical intelligence of robots. Practical tutorials include an introduction to the UniTree Go2 quadruped and exercises on deriving equations of motion using libraries and understanding URDFs.
Block 2: Robots as Dynamical Systems delves into the continuous and hybrid dynamical systems that govern robotic motion. Lectures explore state space forms, phase portraits, and discontinuity. Tutorials provide hands-on experience with energy shaping control of pendulum, passive dynamic walkers, and the SLIP model for walking and running.
In Block 3: Deliberative Planning, the focus shifts to dynamic programming and trajectory optimization. Students learn about the Hamilton Jacobi Bellman Equation, value iteration, and various methods of trajectory optimization. Practical sessions include grid problems, car braking scenarios, and designing brachiation motion with AcroMonk robot.
Block 4: Reactive Control addresses instantaneous and horizon-based stabilization techniques. Lectures cover linear quadratic regulators, task space inverse dynamics, quadratic programming, and Lyapunov analysis. Tutorials involve exercises with simple and double pendulums.
Finally, Block 5: Advanced Topics/Outlook explores cutting-edge topics such as Sim2Real transfer, state estimation, reinforcement learning, and co-design. Lectures are complemented by tutorials on Kalman filters, reinforcement learning applications, and the co-design of robotic components. The block concludes with an outlook on the connection between athletic and cognitive intelligence, including discussions on RRTs and symbolic AI.
Throughout the course, students benefit from consultancy sessions that provide personalized guidance and support. Additionally, there are group assignments designed to encourage collaborative learning and practical application of the concepts covered in the lectures and tutorials.

Literature

Recommended Texts:
  • Underactuated Robotics: Algorithms for Walking, Running, Swimming, Flying, and Manipulation, Russ Tedrake, MIT 2023. (Open access, available online)
  • Modern Robotics: Mechanics, Planning, and Control, Kevin M. Lynch and Frank C. Park, Cambridge University Press, 2017. (Open access, available online)
  • Practical Methods for Optimal Control and Estimation Using Nonlinear Programming, Second Edition, John Betts, SIAM. (available as e-book via Chalmers library)
  • Reinforcement Learning and Optimal Control, Athena Scientific, 2019 (Open access, available online)

Examination including compulsory elements

Completion of all five group assignments is required to pass the course with a grade of 3. Grades of 4 or 5 will be awarded based on performance in the oral examination.

The course examiner may assess individual students in other ways than what is stated above if there are special reasons for doing so, for example if a student has a decision from Chalmers about disability study support.