Make self-driving vehicles dependable

Title of the research project

PERIOD - Pursuing Efficient Reliability of Object Detection for automotive and aerospace applications 

Scientific area

Computer Engineering, Artificial Intelligence, Embedded Systems

Marie Skłodowska-Curie fellow

Paolo Rech


Autonomous vehicles are the next big thing happening in the automotive and space exploration markets. The goal of PERIOD is to design more dependable devices and algorithms for autonomous driving, based on neural networks, in order to avoid the occurrence of dangerous accidents.

Description of the research project 

PERIOD goal is to make the software and hardware frameworks, necessary to implement self-driving vehicles, dependable. These systems are powered by neural networks, that have the role of identifying object in real time and, consequently, take the proper action. This research is not limited to automotive applications, but includes also deep space exploration, thanks to the well-established collaborations with the NASA Jet Propulsion Laboratory (United States), the European Space Agency and the Rutherford Appleton Laboratory (United Kingdom).

Current systems failures can be caused by intrinsic limitations or by external factors, such as interferences, voltage spikes, or the interaction with ionizing particles. Missing an object or detecting an inexistent object are among the most common and risky failures as demonstrated by recent accidents involving self-driving cars.

PERIOD research starts with the understanding of the main causes (hardware and software) of failures, thanks to experimental studies that include the use of particles accelerators to simulate terrestrial and space radiation. Once the computational failures main causes and effects are identified, novel solutions will be designed to increase the dependability of neural networks and of parallel devices required to process frames in real time. The challenge of PERIOD is to reduce as much as possible the computational (and production) overhead required to make autonomous systems dependable.

Impact on fellow career and on society 

Autonomous vehicles, once sufficiently dependable and adopted in large scale, will reduce the number of car accident (which is the forth cause of death) by 3-4 orders of magnitude. Space exploration will have a burst when using autonomous systems, allowing to save the long time required to send signals to deep space. The fellow will bring to this project 10 years of international experience in the field of reliability and dependability.  Moreover, PERIOD will have a great positive impact on his career, allowing at the same time the development of a research field, highly complementary with the work carried on by other groups at Politecnico of Torino

Short CV of Marie Skłodowska-Curie fellow

Paolo Rech received his master and PhD at Padova University in 2006 and 2010, respectively. Since 2012 he is an associate professor at the Universidade Federal do Rio Grande do Sul, Brazil. In 2019 he was the Rosen Scholar Fellow at the Los Alamos National Laboratory (LANL) and in 2020 he received the “Impact on society award” from the Rutherford Appleton Laboratory (RAL). Paolo has active collaborations with industries as ARM, NVIDIA, AMD, Xilinx, with the research labs LANL and RAL, and with the space agencies in the US (JPL) and Europe (ESA).


Matteo Sonza Reorda, DAUIN-Department of Control and Computer Engineering

PERIOD project has received funding from the European Union’s Horizon 2020 research and innovation programme under grant agreement No. 886202

  • Budget: 171.500 euro
  • Start date: 16/11/2020
  • End date: 15/11/2022