Mobile robots are able to localise themselves by sensing their relative position in a static environment. However, when this environment is highly dynamic and heavily occluded, as in the case of a robot moving through a crowd, traditional sensing and localisation techniques break down.
The goal of this project is to design and integrate a custom sensor suite for Pepper, the humanoid robot developed by SoftBank Robotics, such that it can perform robust localisation and navigation in challenging crowded environments. This will involve a system-level exploration and optimisation over the possible combinations of sensing modalities that also leveraging the multiple degrees of freedom available on the Pepper platform (articulated arms, head, torso, etc.)
Mobile robots are able to localise themselves by sensing their relative position in a static environment. However, when this environment is highly dynamic and heavily occluded, as in the case of a robot moving through a crowd, traditional sensing and localisation techniques break down.
The goal of this project is to design and integrate a custom sensor suite for Pepper, the humanoid robot developed by SoftBank Robotics, such that it can perform robust localisation and navigation in challenging crowded environments. This will involve a system-level exploration and optimisation over the possible combinations of sensing modalities that also leveraging the multiple degrees of freedom available on the Pepper platform (articulated arms, head, torso, etc.)
1. Literature review of multi-modality sensing
2. Characterisation of the robot platform within the sensing environment
3. Design and implementation of a sensor suite for navigation through crowds
4. Sensor calibration and integration with the existing software/hardware on Pepper
5. Experimentation to evaluate and validate the performance of the complete sensor package
1. Literature review of multi-modality sensing 2. Characterisation of the robot platform within the sensing environment 3. Design and implementation of a sensor suite for navigation through crowds 4. Sensor calibration and integration with the existing software/hardware on Pepper 5. Experimentation to evaluate and validate the performance of the complete sensor package
- Strong interest in working on robotic hardware
- Programming skills (C++/Python)
- Experience with Linux, ROS, and typical development tools such as Git
- Strong interest in working on robotic hardware - Programming skills (C++/Python) - Experience with Linux, ROS, and typical development tools such as Git
- Jen Jen Chung, jenjen.chung@mavt.ethz.ch
- Mark Pfeiffer, mark.pfeiffer@mavt.ethz.ch
- Jen Jen Chung, jenjen.chung@mavt.ethz.ch - Mark Pfeiffer, mark.pfeiffer@mavt.ethz.ch