If you are interested in one of the below projects, Semester Projects, Master Projects at EPFL, or Master Projects in Industry please contact the first person of reference indicated in each description either by telephone, or by email, or by visiting us directly to the LASA offices.
Shared control of robot with EEG signals in virtual and real environments
Motivation: This semester project would focus on the use brain-computer interfacing (BCI) for assisting in robot control by humans. We would evaluate the use of electroencephalogram (EEG) signals from human brain as a feedback to correct robot’s behavior. In particular, EEG signals can exhibit a special characteristic called Error Related Potential (ErRP) whenever the subject observes a potential accident or anomalous incident. Machine learning (ML) algorithms are commonly deployed to detect ErRP by decoding EEG signals. At LASA lab, we have conceived robotic experimental setups where ErRP signals can be used in inverse learning algorithms to improve robot’s trajectory. The student is expected to contribute to the ongoing experimental work in this direction during the semester.
Description: The robotic setup consists of a standing wheelchair robot, named as “Qolo” designed for patients with spinal cord disability. Using dynamical systems based obstacle avoidance algorithm, the robot with subject riding on it can navigate in a given environment. The subject, in turn wears a cap with EEG sensors embedded in it. This would allow us to detect an ErRP in the subject’s EEG signals before a potential collision and help the IRL algorithm possibly avoid it. To decode EEG signals first, a ML based classifier would be trained while simulating the experiments virtually. The testing phase would subsequently be executed on the real robot to evaluate the performance of the ML algorithm and IRL algorithm. The student would primarily contribute in assisting these experiments for collecting data but would also have the opportunity to design or improve the ML based decoder.
|Section(s)||ME MT SV EL IN MA|
|Type||50 % Experiments, 30 % software, 20 % Analysis|
|Knowledge(s)||(Desirable but not required): Python, ROS, Unity, IRL|
|Subjects(s)||Brain-Computer Interfacing, Robotics, Virtual Reality|
Human-Robot Collision Simulation and Experimental Validation
During this semester project, we propose an experimental robotics application for assessing the safety of mobile robots around human pedestrians. The main objective is to develop a human-robot collision simulation model validated with data. You will benefit from access to multiple mobile robots (virtually and physically), learning to control them with different motion controllers (Velocity control, Impedance control)for evaluating collision data under current simulation framework  [Link to code]. Moreover, we will assess the collision between a robot and fixed objects in multiple scenarios in the virtual platform for matching the given data set. Many elements can affect the robot’s dynamic response to collisions, e.g. the robot’s mechanical composition (weight, structural stiffness), its actuators, and its wheels’ interaction with the ground(stick-slip behaviour). The scientific literature about mobile robots still lacks general treatments of these properties’ effect on robots’ response to collisions. The student is expected to have good experience in python and Matlab, solid knowledge of dynamical systems modelling, and a good understanding of solid mechanics and deformation analysis would be a plus.
The model’s development will target the robot Qolo , a powered standing wheelchair, which is equipped with a bumper and force sensor for collision experiments. The work is divided into the following steps:
1. Defining the robot’s dynamic model as a set of inputs and outputs matching available data.
2. Implementing the dynamic model under python (pybullet) and evaluating collision responses.
3. (Optional) Evaluating the parametric model for the system from physical principles of collision after reviewing the literature and assessing available data.
|Section(s)||EL IN MA ME MT PH|
|Type||20% theory, 30% software, 50% implementation|
|Knowledge(s)||C++, Python/MATLAB, Robotics, Mechanics|
|Subjects(s)||Human-Robot Interaction, Mechanis|
|Responsible(s)||Diego F. Paez Granados, David Gonon|
Master Projects at EPFL
No available projects.