Biorobotics Laboratory BioRob

Project Database

This page contains the database of possible research projects for master and bachelor students in the Biorobotics Laboratory (BioRob). Visiting students are also welcome to join BioRob, but it should be noted that no funding is offered for those projects. To enroll for a project, please directly contact one of the assistants (directly in his/her office, by phone or by mail). Spontaneous propositions for projects are also welcome, if they are related to the research topics of BioRob, see the BioRob Research pages and the results of previous student projects.

Amphibious robotics
Computational Neuroscience
Dynamical systems
Human-exoskeleton dynamics and control
Humanoid robotics
Miscellaneous
Mobile robotics
Modular robotics
Neuro-muscular modelling
Quadruped robotics


Amphibious robotics

710 – Data gathering and analysis with DeepLabCut for robots.
Category:semester project
Keywords:Bio-inspiration, Data Evaluation, Data Processing, Image Processing, Learning, Python, Robotics
Type:20% theory, 40% hardware, 40% software
Responsible: (MED 1 1611, phone: 36620)
Description:Designing experiments with streamlined protocols is essential for gathering data and analysing the behaviour. This becomes especially relevant for comparing data between animals and robots. The work in this project would involve using hardware and software tools for high-quality data gathering and analysis. For instance setting up cameras, performing calibration, and exploring tools such as DeepLabCut for tracking robots with an articulated spine. If needed, develop new algorithms for tracking the joints of the robot. Performing kinematic analysis & computing evaluation metrics. Skills gained: Experiment design & setup, Working with cameras or tracking frameworks to gather data, python, exposure to bio-inspired controllers (central pattern generators), hands-on experience with bio-inspired robots, using state-of-the-art tools for analysis. To apply, please email your CV Astha to with a short paragraph of relevant courses and skills that can contribute to the project.

Last edited: 01/02/2023
702 – Design and control of a swimming soft robot with variable stiffness
Category:master project (full-time)
Keywords:Compliance, Control, Experiments, Robotics
Type:20% theory, 60% hardware, 20% software
Responsible: (MED 1 1226, phone: 32658)
Description:

Robots still lag behind their natural counterparts despite the recent proliferation of computation, sensing, and actuation at a small scale. Much of this observed performance deficiency stems from differences in structure, particularly the use of compliant structures with distributed strain sensing and mechanical feedback.

Physiological experiments have shown that in fish, significant energetic efficiency is derived by exploiting the stiffness of their body structure, converting energy from the fluid to the body and resulting in propulsion being generated passively. The internal dynamics and the mechanisms through which soft structures are exploited for attaining this remarkable energy efficiency are underexplored. The interplay between active and passive stiffness control is the subject of fundamental science and offers significant potential for biomimetic technology transfer.

This project aims to understand how active modulation of body stiffness across swimming modes and fin shapes affects swimming speed and maximum thrust.

A biologically realistic robophysical model will be designed and tested to accomplish that. An integration of a soft pressure sensor will be explored and closed-loop swimming control based on this sensor information investigated. Performance will be studied by comparing open loop vs. closed loop control with experimental validation.

Investigating the underlying mechanisms present in soft structures in moving fluids will enhance robot mobility and offer important insights in neuromechanics, by creating robots that can function as ’model animals’ for biomechanics research.

This work will be done in a collaboration between Dr. Ardian Jusufi at EMPA and the Biorobotics lab at EPFL. It will involve some travel between the two places.



Last edited: 09/09/2022
675 – How to achieve good visual motor coordination during fast movements?
Category:semester project, master project (full-time), internship
Keywords:Bio-inspiration, C++, Data Processing, Experiments, Image Processing, Optic Flow, Programming
Type:10% theory, 15% hardware, 75% software
Responsible: (MED 1 1023, phone: 31367)
Description:To survive in nature, an animal has to achieve both fast movements and also maintain good Sensory-motor coordination (e.g. visual-motor coordination) in the meantime. How to achieve this? Uncovering its underlying theory can help understand the intelligence and make a better robot. Zebrafish larvae swim at a high tail-beating frequency. Such movement induces strong disturbance on the visual sensation. Recent scientific findings on zebrafish larvae revealed the slow and delayed response of neurons in the brain. We hypothesize the properties of such neurons help deal with such a challenge. In this project, a student will conduct simulations of a swimming robot (e.g. Envirobot in Biorob) in Webots simulator and assist the supervisors to answer this question. (This project targets the students who want to continue their study in academia or are willing to stay longer even after the semester, For more details, please contact the supervisors.) To do the work, the student needs a good ability in programing language (e.g. C and C++). Interests in biology and neuroscience would be favored. Also, this project requires a long-term enrollment of the student (this means the student has to keep his/her connection with this project after the end of the semester, although there would not be much work).

Last edited: 16/08/2022

Human-exoskeleton dynamics and control

708 – Investigation of a bio-inspired assistive controller for hip exoskeletons
Category:master project (full-time), internship
Keywords:Bio-inspiration, Biped Locomotion, C++, Control, Experiments, Robotics
Type:30% theory, 30% hardware, 40% software
Responsible: (MED 3 1017, phone: 35825)
Description:One of the applications of lower-limb exoskeletons is to make walking easier and less effortful for people, and thus to augment the endurance and mobility of their users, similar to what e-bikes can do for cyclists. This requires cooperative interaction of the robot with the human, by correctly predicting the human's intended movement and providing appropriate assistive forces at the right time. Achieving such a synergistic interaction is a challenging task and many different control algorithms have been developed to address this challenge. One bio-inspired approach is the so-called NeuroMuscular Controller (NMC), which is inspired by the reflexes of the human neuro-muscular system.

A version of NMC has already been implemented on an existing hip exoskeleton in Biorob's REHAssist group, and preliminary tests with human subjects have shown promising early results. However, there are still some minor improvements to be done on the controller, and more in-depth experimentation is needed to study its performance. The goal of this project is to (i) continue the development of this controller, and (ii) to carry out in-depth and systematic experiments with human subjects to study how the assistance provided by the NMC affects the energy consumption during walking.

The ideal candidate should have good C++ programming skills (familiarity with classes and polymorphism is highly recommended), and at least some experience with embedded systems and programming for hardware implementation (only to use the existing hardware and troubleshoot if needed, no hardware development is required). Basic signal processing and analysis skills (MATLAB/Python) are also helpful. Familiarity with human biomechanics and gait analysis is a plus, but not required. This project will very likely result in a scientific publication, so it can be particularly interesting to students interested in a research-oriented career or a PhD.

In case of interest, please contact the supervisor with a copy of your CV and a short description of your relevant experience.



Last edited: 03/01/2023

Quadruped robotics

A small excerpt of possible projects is listed here. Highly interested students may also propose projects, or continue an existing topic.

707 – Sample-Efficient Locomotion Learning for Quadruped Robots
Category:semester project, master project (full-time), internship
Keywords:Learning, Locomotion, Optimization, Programming, Python, Quadruped Locomotion
Type:25% theory, 25% hardware, 50% software
Responsible: (MED 1 1611, phone: 36714)
Description:Animals learn to walk much faster and more efficient than legged robots in the context of agile and robust locomotion. In this project, the student will develop sample-efficient and fast locomotion learning based on Bayesian Optimization (BO) and Central Pattern Generator (CPG). The student will use the Python libraries for Bayesian Optimization to optimize the CPG parameters for quadruped locomotion gaits. Initially, the implementation will be performed in a simulator such as PyBullet, and then it will be transferred to the real robot. The experience in python programming and legged robotics are required for the project. Application: To apply, please email your CV, a code sample, and your motivation to milad.shafiee@epfl.ch

Last edited: 16/12/2022
667 – Simulation and Characterization of a bio-inspired multi-segmented leg consisting of spring and tendon
Category:semester project, master project (full-time), internship
Keywords:C++, Compliance, Dynamics Model, Leg design, Locomotion, Programming, Python
Type:50% theory, 50% software
Responsible: (MED 1 1611, phone: 36714)
Description:The development of agile-legged robots has attracted significant attention, yet there are few robots that mimic the structures and functions of biological muscles and tendons of quadruped mammals. The driving force of muscles and the elastic force of muscles and tendons can help improve the performance and efficiency of quadruped robots. The goal of this project is to take a step toward bridging the biology and robotics in leg mechanisms of quadrupeds. In this project, the student will design and perform dynamic characterization of the leg by considering the different configurations of spring and tendon. In this way, the student will use a CAD tool (i.e. inventor or Solidworks) to mechanically design a three-segment robot leg consisting of spring and tendon. Then, the student will analyze and characterize the designed leg in simulation for a dynamic task such as jumping in place by using dynamics simulators (i.e Mujoco). For this project, the student will need to be familiar with dynamics concepts, CAD tools, and python programming. For more information please send your CV and sample of related projects that you have already conducted.

Last edited: 12/12/2022
652 – Integrating Learning-Based Control with MPC and CPGs
Category:semester project, master project (full-time), internship
Keywords:Bio-inspiration, Control, Learning, Locomotion, Optimization, Robotics
Type:40% theory, 60% software
Responsible: (MED 1 1024, phone: 37506)
Description:Recent years have shown impressive locomotion control of dynamic systems through a variety of methods, for example with optimal control (MPC), machine learning (deep reinforcement learning), and bio-inspired approaches (CPGs). Given a system for which two or more of these methods exist: how should we choose which to use at run time? Should this depend on environmental factors, i.e. the expected value of a given state? Can this help with explainability of what exactly our deep reinforcement learning policy has learned? In this project, the student will use machine learning to answer these questions, as well as integrate CPGs and MPC into the deep reinforcement learning framework. The methods will be validated on systems including quadrupeds and model cars first in simulation, with the goal of transferring the method to hardware.

Last edited: 10/12/2022

Miscellaneous

703 – Resurrect Dinosaur with Soft Robot Modeling
Category:master project (full-time)
Keywords:Artificial muscles, Biomimicry, Machine learning, Prototyping
Type:100% other activities
Responsible: (MED 1 1226, phone: 32658)
Description:

Robots still lag behind their natural counterparts despite the recent proliferation of computation, sensing, and actuation at a small scale. Much of this observed performance deficiency stems from differences in structure, particularly the use of compliant structures with distributed strain sensing and mechanical feedback.

Weaponry for intraspecific combat or predator defense is one of the most common animal adaptations, but the selective pressures and constraints governing its phenotypic diversity and skeletal regionalization are poorly understood. The evolution of tail weaponry in amniotes, a rare form of weaponry that nevertheless evolved independently in a wide variety of organisms, including mammals, turtles, and dinosaurs has been studied using phylogenetic comparative techniques (Arbour and Zanno 2018). Especially the relationships between morphology, ecology, and behavior in extant animals known to use the tail as a weapon were researched.

Because large, armored herbivores are so rare in today's terrestrial faunas, we propose that evolution of tail weaponry is prone to be investigated with bio-inspired robotic platforms, offering us the potential to investigate paleontoloigcal evolution of tails. The interplay between tail length, stiffness, and other traitsl is the subject of fundamental science and offers significant potential for biomimetic technology transfer.

This project aims to understand how active modulation of body stiffness across swimming modes and fin shapes affects swimming speed and maximum thrust.

A biologically realistic robophysical model will be designed and tested to accomplish this project. An integration of a soft pressure sensor will be explored and tendon-driven closed-loop control based on this sensor information investigated. Performance will be studied by comparing open loop vs closed loop control with experimental validation.

Investigating the underlying mechanisms present in soft structures in tails will enhance robotic mobility and offer important insights in neuromechanics, by creating robots that can function as ’model animals’ for biomechanics research.

This work will be done in a collaboration between Dr Ardian Jusufi at EMPA and the Biorobotics lab at EPFL. It will involve some travel between the two places.

Your task

  • Build robotic tail platform
  • Improve control method of the tail club to strike a target accurately
  • fabricate pressure-sensitive tactile sensing that is strain invariant
  • Integrate sensing capability into soft robophysical platform
  • Explore innovative applications.

You have…

  • Mechatronics background
  • Mechanical design
  • Knowledge in machining, fabrication and materials, as it pertains to soft fluidic actuators
  • Creative and hands-on skills
  • Highly motivated and independent
  • Experienced in CAD


Last edited: 10/11/2022

Neuro-muscular modelling

704 – Artificial Intelligence and Biomechanical Modelling to optimize motor patterns during Epidural Electrical Stimulation
Category:master project (full-time)
Keywords:Computational Neuroscience, Machine learning, Muscle modeling, Programming, Python
Type:30% theory, 70% software
Responsible: (MED 1 1611, phone: 36714)
Description:Spinal cord injury disrupts the communication between the brain and the spinal circuits below the lesion that generate and coordinate limb movements. Epidural Electrical Stimulation (EES) can restore movements of paralyzed limbs. The effects of EES vary depending on frequency, amplitude and location. To effectively restore walking, therapists need to calibrate a spectrum of stimulation protocols. Even for a simple functional movement such as walking, the calibration is performed over several weeks by expert engineers. This process presents a barrier for real-life worldwide use of the therapy. In this Master thesis, we propose to leverage concepts of Artificial Intelligence (AI) and Biomechanical Modelling to better understand and predict EES-induced motor patterns. To do so, we will make use of a personalized biomechanical model of the lower body and recorded kinematic data of a treated participant with EES. We will then train AI architectures to simulate the Biomechanical and reproduce the recorded kinematic data. Doing so, we will learn the relationships between EES and motor patterns. Our aim is to identify a spectrum of EES stimulation protocols that generate desired functional movements and validate it during clinical trials. During this Master thesis, the student will learn concepts of Artificial Intelligence, Data Analysis and Biomechanical Modelling.

Last edited: 21/11/2022

Mobile robotics

651 – Autonomous Drifting on Scaled Vehicle Hardware
Category:semester project, master project (full-time), internship
Keywords:C++, Control, Electronics, Embedded Systems, Experiments, Learning, Optimization
Type:10% theory, 60% hardware, 30% software
Responsibles: (MED 1 1024, phone: 37506)
(ME C2 374, phone: 33190)
Description:Controlling vehicles at their limits of handling has significant implications from both safety and autonomous racing perspectives. For example, in icy conditions, skidding may occur unintentionally, making it desirable to safely control the vehicle back to its nominal working conditions. From a racing perspective, drivers of rally cars drift around turns while maintaining high speeds on loose gravel or dirt tracks. In this project, the student will compare several approaches for high speed, dynamic vehicle maneuvers, including NMPC with a standard dynamic bicycle model, NMPC with a dynamic bicycle model + GP residuals, NMPC with learned dynamics (i.e. a NN), and lastly a pure model-free reinforcement learning approach. All approaches will be tested in both simulation as well as on a scaled vehicle hardware platform.

Last edited: 10/12/2022

10 projects found.