DISAL offers a range of student projects its three main areas of expertise: Distributed Robotic Systems, Sensor and Actuator Networks, and Intelligent Vehicles. For more information about supervision guidelines and different types of student projects please refer to this page.
Fall Semester 2023-2024
Using Artificial Neural Networks for the Automatic Design of Behavioral Arbitrators for Khepera IV Robots
Assigned to: Peter Harmouch
![]() |
![]() |
With the increased deployment of mobile robotics in a multitude of real-life scenarios, robots must face more and more complex situations. However, every robotic platform needs to be carefully programmed to fulfill its task, which, today, is often still done manually. Such programming operation usually represents an intense time investment and requires high expertise to eventually achieve the targeted performances in a real deployment, even more so if the application involves a multi-robot system. There are multiple approaches for the automatic generation of controllers (or specific components), whereas the nature of the control components ranges from Probabilistic Finite State Machines (PFSM) [1] to Artificial Neural Networks (ANNs) [2]. However, powerful machine-learning techniques are necessary to automatically generate any of these controllers, requiring numerous, computationally expensive simulations.
This project aims to leverage an ANN to replace computationally expensive high-fidelity simulations during the automatic design of robotic control arbitrators.
Starting from the results of an exploratory project which used graph neural networks, this project aims to design, train and evaluate an ANN able to accurately model the behavior of different basic behaviors, which can be seen as the building blocks of an FSM-based arbitrator controller [1]. After an initial validation of the model’s performance with respect to the ground truth, obtained using the Webots high-fidelity robotics simulator, the model will then be leveraged to automatically design an FSM-based arbitrator for Khepera IV robots. The expected effort and outcome of this work will be adjusted based on the type of project.
Recommended type of project: Master project / Semester project
Work breakdown: 30% theory, 50% software, 20% experimentation
Prerequisites: broad interest in machine learning, neural networks, control algorithms and robotics
Keywords: machine learning, robot control, automatic design and optimization
Contact: Cyrill Baumann
Reference:
[1] C. Baumann, H. Birch, A. Martinoli, “Leveraging Multi-Level Modelling to Automatically Design Behavioral Arbitrators in Robotic Controllers”, IEEE/RSJ International Conference on Intelligent Robots and Systems, 2022, pp. 9318-9325.
[2] Duarte M., Sancho Moura O., and Anders Lyhne C., “Evolution of Hybrid Robotic Controllers for Complex Tasks” Journal of Intelligent & Robotic Systems, 2015, 78: 463
Point cloud segmentation of infrastructural steel elements
Assigned to: Dimitri Jacquemont
![]() |
![]() |
Community-critical steel infrastructure, such as energy transmission and telecommunication lines, steel bridges, offshore rigs, and so on, requires regular inspection in order to guarantee the integrity of the structure and good operation. Current inspection approaches usually require the use of heavy equipment, as well as visual inspection performed by technicians, which exposes them to potentially dangerous situations. Furthermore, inspections usually imply downtime for the structure under inspection, which is costly and inefficient. The main interest is, therefore, to identify fatigue signs within a structure without requiring to stop its usage. Micro Aerial Vehicles (MAVs) endowed with visual and depth sensing capabilities are potentially good candidates to achieve this goal.
A previous project already explored different techniques [1], allowing to segment point clouds to understand the geometrical component of a monitored steel infrastructure. It relies on standard tools and libraries (such as PCL [3] and Open3D [4]) and an Iterative Hough Transform for Line Detection in 3D Point Clouds [5]. However, real-time segmentation is not guaranteed to this point. This project will therefore aim to address some shortcomings of the current solution to allow for real-time operation. This will include the implementation of efficient point cloud resampling (down-sampling, filtering, voxelization [2]) and pre-processing algorithms for relevant feature extraction and proper evaluation of the performance (profiling). Additional functionalities and optimizations can be implemented (e.g., a deep learning-based approach). The preferred programming language for this project is C++, in combination with ROS ideally. Tests in simulation leveraging the Webots simulator [6] and on a real MAV equipped with a ToF camera will be carried out if time permits. In any case, real point cloud data are available or can be generated and should be used for validation.
Recommended type of project: semester project
Work breakdown: 30% theory, 70% simulation
Prerequisites: Broad interest in robotics, very good programming skills (C/C++, Python). Knowledge in ROS, Webots, git. Having attended a computer vision class is recommended.
Keywords: Webots, simulation, MAV, point cloud, 3D, segmentation.
Contact: Lucas Wälti
References:
[1] Nguyen, Anh, and Bac Le. “3D Point Cloud Segmentation: A Survey.” In 2013 6th IEEE Conference on Robotics, Automation and Mechatronics (RAM), 225–30, 2013. https://doi.org/10.1109/RAM.2013.6758588.
[2] Oleynikova, Helen, Zachary Taylor, Marius Fehr, Roland Siegwart, and Juan Nieto. “Voxblox: Incremental 3D Euclidean Signed Distance Fields for on-Board MAV Planning.” In 2017 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), 1366–73, 2017. https://doi.org/10.1109/IROS.2017.8202315.
[3] Point Cloud Library (PCL): https://pointclouds.org/
[4] Zhou, Qian-Yi, Jaesik Park, and Vladlen Koltun. “Open3D: A Modern Library for 3D Data Processing.” arXiv, January 29, 2018. https://doi.org/10.48550/arXiv.1801.09847.
[5] Dalitz, Christoph, Tilman Schramke, and Manuel Jeltsch. “Iterative Hough Transform for Line Detection in 3D Point Clouds.” Image Processing On Line 7 (July 19, 2017): 184–96. https://doi.org/10.5201/ipol.2017.208.
[6] “Webots: Robot Simulator.” Accessed October 6, 2022. https://cyberbotics.com/.
Local navigation for the inspection of steel structures
Assigned to: Louis Marchand

Community-critical steel infrastructure, such as energy transmission and telecommunication lines, steel bridges, offshore rigs, and so on require regular inspection, in order to guarantee the integrity of the structure and good operation. Current inspection approaches usually require the use of heavy equipment, as well as visual inspection performed by technicians, which exposes them to potentially dangerous situations. Furthermore, inspections usually imply down-time for the structure under inspection, which is costly and inefficient. The main point of interest is therefore to identify fatigue signs within a structure, without requiring to stop its usage. Micro Aerial Vehicles (MAVs) are good candidates to achieve this goal.
This project aims at implementing a navigation solution for an autonomous MAV, enabling it to perform smooth reliable obstacle avoidance maneuvers when navigating between waypoints in a cluttered environment [5]. The MAV is equipped with a forward looking ToF camera, which allows it to reliably detect obstacles. A previous replanning solution was implemented, however, some shortcomings persist. This project will therefore aim at verifying the current solution, implement a visualization of the collision resolution process along the trajectory, and a possible re-implementation could be performed if deemed suitable (following a similar or new approach). The obstacle map is stored using Voxblox [1][2] and polynomial splines are used to describe the trajectory [3]. This project will mostly rely on ROS and C++. Simulations in Webots [4] will be used for developing and testing the solution.
Recommended type of project: semester project
Work breakdown: 50% theory, 50% simulation
Prerequisites: Broad interest in robotics, very good programming skills (C/C++, Python). Knowledge in control theory, ROS, Webots, git. Having attended the course “Aerial robotics” or similar is recommended.
Keywords: Webots, simulation, quadrotors, auto-pilot, flight-controller, computer vision, navigation, path planning, ROS.
Contact: Lucas Wälti
References:
[1] Oleynikova, Helen, Zachary Taylor, Marius Fehr, Roland Siegwart, and Juan Nieto. “Voxblox: Incremental 3D Euclidean Signed Distance Fields for on-Board MAV Planning.” In 2017 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), 1366–73, 2017. https://doi.org/10.1109/IROS.2017.8202315.
[2] Oleynikova, Helen, Michael Burri, Zachary Taylor, Juan Nieto, Roland Siegwart, and Enric Galceran. “Continuous-Time Trajectory Optimization for Online UAV Replanning.” In 2016 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), 5332–39, 2016. https://doi.org/10.1109/IROS.2016.7759784.
[3] Richter, Charles, Adam Bry, and Nicholas Roy. “Polynomial Trajectory Planning for Aggressive Quadrotor Flight in Dense Indoor Environments.” In Robotics Research, edited by Masayuki Inaba and Peter Corke, 114:649–66. Springer Tracts in Advanced Robotics. Cham: Springer International Publishing, 2016. https://doi.org/10.1007/978-3-319-28872-7_37.
[4] Webots: https://cyberbotics.com/doc/reference/index
[5] Yang, Liang, Juntong Qi, Jizhong Xiao, and Xia Yong. “A Literature Review of UAV 3D Path Planning.” In Proceeding of the 11th World Congress on Intelligent Control and Automation, 2376–81, 2014. https://doi.org/10.1109/WCICA.2014.7053093.
Wind Sensing for Miniature Quadrotors: Design and Validation
Assigned to: Raphaël Dousson
![]() |
![]() |
Wind sensing on a miniature mobile robotics platform might be very challenging because of severe volume, mass, or power constraints. For instance, on a rotor-winged Micro Aerial Vehicles (MAV) such challenges are exacerbated by limited payload and agile movements of the platform. Traditional anemometers, such as mechanical and ultrasonic ones, are often too bulky to be carried by MAVs. On the other hand, temperature- and pressure-based anemometers are technologically more promising to obtain customized solutions for MAVs. At DISAL, we have developed a prototype of a temperature-based wind sensor module designed for wheeled mobile robots. The module utilizes six thermistors placed inside 3D printed tubes arranged in a star-like configuration, where the differences among thermistor readings are used to deduce the wind direction with an accuracy smaller than ten degrees. While promising, this technology needs to be further evaluated using existing reference instruments, possibly expanded to full anemometric measurement (e.g., including wind strength), and investigated for possible 3D measurements.
First, a systematic calibration and validation process of the existing wind sensor module will be conducted using a real wheeled robot and dedicated wind tools (e.g., fan, wind tunnel, reference anemometers). Second, a new prototype (including firmware and mechatronic design) will have to be produced to upgrade the capabilities of the existing sensor from a given perspective (e.g., reduce its volume/mass, measuring wind strength, measuring in 3D).
Recommended type of project: semester project, master project
Work breakdown: 70% programming, 30% mechatronics design
Prerequisites: passion on solving issues, broad interest in robotics and sensors, good programming skill (C/C++), experience with real robots will be an asset
Keywords: wind sensing, sensors, embedded systems
Contact: Wanting Jin
References:
[1] H. Ishida, H. Tanaka, H. Taniguchi, and T. Moriizumi, “Mobile robot navigation using vision and olfaction to search for a gas/odor source,” Autonomous Robots, vol. 20, no. 3, pp. 231238, June 2006
[2] T. Lochmatter, “Bio-inspired and probabilistic algorithms for distributed odor source localization using mobile robots,” Ph.D. dissertation, EPFL, Lausanne, 01 2010. Chapter 2
Sensing Systems for Gas Source Localization: A Comparative Study
Assigned to: Duranceau Agatha Ludivine
![]() |
![]() |
Gas Source Localization (GSL) is a challenging task in the real world due to the chaotic nature of the gas dispersion phenomenon, which results in intermittent and noisy gas measurements. Sensing assets characterized by different mobility capabilities can be used to carry out GSL (e.g., static sensor network and mobile robots). Each asset shows strengths and weaknesses. For instance, a sensor network can gather gas measurements at different positions simultaneously over a long-time horizon but they cannot refine its deployment over time. On the other hand, mobile robotic platforms can plan their motion dynamically with respect to the current perception and estimation of the environment. However, the measurement quality is influenced by the robot’s mobility, making the measurements less reliable and limited in time.
This project aims to investigate the benefits of both solutions for carrying out GSL in the very same controlled scenario, reproduceable both in high-fidelity simulation and a in a wind tunnel. In particular, we are interested in understanding what plume features should be extracted from the gas readings to optimize performances of GSL algorithms (e.g., localization accuracy, convergence time) [2][3]. The student will be able to start from multiple baseline probabilistic algorithmic solutions for carrying out the comparative study.
Recommended type of project: semester project, master project
Work breakdown: 60% theory, 40% simulation
Prerequisites: passion on solving issues, broad interest in robotics, good programming skill (Python/C++), experience with Webots will be an asset
Keywords: gas source localization, heterogenous system, sensor fusion
Contact: Wanting Jin
References:
[1] W. C. Evans, D. Dias, S. Roelofsen, and A. Martinoli, “Environmental field estimation with hybrid-mobility sensor networks,” in 2016 IEEE International Conference on Robotics and Automation (ICRA). Stockholm: IEEE, May 2016, pp. 5301–5308.
[2] J. G. Li, Q. H. Meng, Y. Wang, and M. Zeng, “Odor source localization using a mobile robot in outdoor airflow environments with a particle filter algorithm,” Auton. Robot., vol. 30, no. 3, pp. 281–292, 2011.
[3] J. Burgues, V. Hernandez, A. J. Lilienthal, and S. Marco, “Gas ´ distribution mapping and source localization using a 3D grid of metal oxide semiconductor sensors,” Sensors and Actuators B: Chemical, vol. 304, p. 127309, Feb. 2020.
Spring Semester 2023-2024
The student projects for the next spring semester will be posted at the beginning of November 2023.