IDSIA engages in both applied and basic research in the field of autonomous robotics, often intersecting with other research domains, particularly machine learning and visual computing. The institute is currently involved in many robotics research collaborations with national and international partners, and works with mobile robots (flying and ground), as well as industrial robotics.
Area leaders:
Autonomous Robotics - IDSIA NEW
Scientific area
Autonomous robotics
SUPSI Image Focus
Modern autonomous robots come equipped with advanced sensors like cameras and lidars, generating large volumes of high-dimensional data. To interpret this data and enable autonomous behavior, robots use cutting-edge deep learning techniques. The group focuses on two key research challenges:
- Enabling robots to learn and adapt their perception models autonomously using self-supervised learning techniques.
- Developing applications for novel and challenging tasks for mobile and industrial robots, when limited data is available for training.
Notable past achievements include vision-based control of nano drones near humans, robots learning to detect long-range obstacles autonomously, a quadrotor navigating forest trails, and ground robots estimating traversability on tough terrain. These advancements have also been applied in industrial R&D projects, such as adaptive visual quality inspection systems for various products.
The group focuses its scientific effort on improving the onboard intelligence of ultra-constrained miniaturized robotic platforms aiming at the same capability as biological systems. By leveraging state-of-the-art AI algorithms, the group focuses on optimized ultra-low power embedded Cyber-Physical Systems, deep learning models for energy-efficient perception pipelines, multi-modal ultra-low power sensor fusion, and Cyber-secure systems for Microcontroller Units-class (MCUs).
Thanks to the close partnership with the Parallelel Ultra-low Power international research project (PULP Platform), the group boasts strong collaborations with the ETH Zürich, the University of Bologna, and the Polytechnic University of Torino.
Thanks to the close partnership with the Parallelel Ultra-low Power international research project (PULP Platform), the group boasts strong collaborations with the ETH Zürich, the University of Bologna, and the Polytechnic University of Torino.
Social Robotics and Human-Robot Interaction
Leader: Antonio Paolillo
In the near future, robots will become increasingly present in everyday life, creating a growing demand for easy-to-use, interactive, and adaptive machines designed for non-expert users. Our group leverages state-of-the-art techniques in robot control, machine learning, and AI to develop advanced perception and actuation capabilities for social robots—enabling them to exhibit human-friendly, predictable, and efficient behaviors across a variety of real-world settings.
Collaboration and planning in multi-robot systems
Leader: Jérôme Guzzi
IDSIA is working on the interplay between communication and coordination in multi-agent systems, focusing on mixed groups of robots and humans. The institute investigates different algorithmic strategies where even very minimal communication between the agents favors group coordination: from bio-inspired models to imitation and reinforcement learning. Past research includes artificial emotions for multi-robot coordination and human-friendly robot navigation algorithms. The group validates its results by performing experiments with real robots as well as virtual robots interacting with people in Virtual and Mixed Reality environments.
Robotics for Education
Leader: Luca Gambardella
Using robots in education is a very important interdisciplinary field of research, at the crossing between educational sciences and robotics. The institute has been part of Introducing People to Research in Robotics through an Extended Peer Community in Southern Switzerland: a project awarded the Optimus Agora Prize by the Swiss National Science Foundation.
-
Drones
-
Industrial Robots
-
Service Robots
-
Rescue and Mapping Robots
Self-supervised prediction of the intention to interact with a service robot
G Abbate, A Giusti, V Schmuck, O Celiktutan, A Paolillo
Robotics and Autonomous Systems 171, 104568
A sim-to-real deep learning-based framework for autonomous nano-drone racing
Lorenzo Lamberti, Elia Cereda, Gabriele Abbate, Lorenzo Bellone, Victor Javier Kartsch Morinigo, Michał Barciś, Agata Barciś, Alessandro Giusti, Francesco Conti, Daniele Palossi
IEEE Robotics and Automation Letters 9 (2), 1899-1906
An outlier exposure approach to improve visual anomaly detection performance for mobile robots
D Mantegazza, A Giusti, LM Gambardella, J Guzzi
IEEE Robotics and Automation Letters 7 (4), 11354-11361
Fully onboard ai-powered human-drone pose estimation on ultralow-power autonomous flying nano-uavs
D Palossi, N Zimmerman, A Burrello, F Conti, H Müller, LM Gambardella, ...
IEEE Internet of Things Journal 9 (3), 1913-1929
Learning ground traversability from simulations
RO Chavez-Garcia, J Guzzi, LM Gambardella, A Giusti
IEEE Robotics and Automation letters 3 (3), 1695-1702
A machine learning approach to visual perception of forest trails for mobile robots
A Giusti, J Guzzi, DC Cireşan, FL He, JP Rodríguez, F Fontana, ...
IEEE Robotics and Automation Letters 1 (2), 661-667
G Abbate, A Giusti, V Schmuck, O Celiktutan, A Paolillo
Robotics and Autonomous Systems 171, 104568
A sim-to-real deep learning-based framework for autonomous nano-drone racing
Lorenzo Lamberti, Elia Cereda, Gabriele Abbate, Lorenzo Bellone, Victor Javier Kartsch Morinigo, Michał Barciś, Agata Barciś, Alessandro Giusti, Francesco Conti, Daniele Palossi
IEEE Robotics and Automation Letters 9 (2), 1899-1906
An outlier exposure approach to improve visual anomaly detection performance for mobile robots
D Mantegazza, A Giusti, LM Gambardella, J Guzzi
IEEE Robotics and Automation Letters 7 (4), 11354-11361
Fully onboard ai-powered human-drone pose estimation on ultralow-power autonomous flying nano-uavs
D Palossi, N Zimmerman, A Burrello, F Conti, H Müller, LM Gambardella, ...
IEEE Internet of Things Journal 9 (3), 1913-1929
Learning ground traversability from simulations
RO Chavez-Garcia, J Guzzi, LM Gambardella, A Giusti
IEEE Robotics and Automation letters 3 (3), 1695-1702
A machine learning approach to visual perception of forest trails for mobile robots
A Giusti, J Guzzi, DC Cireşan, FL He, JP Rodríguez, F Fontana, ...
IEEE Robotics and Automation Letters 1 (2), 661-667