Jérémie Guiochet is professor in computer science at University of Toulouse. His research activities take place at LAAS-CNRS. He received an Engineer degree in electronics and computer sciences at ENSEIRB, and a Ph.D. in 2003 from the INSA, Toulouse, France. He provides courses in Computer Science, Industrial Computing, Dependabilty and Risk Management. His research work is conducted in the LAAS-CNRS in the dependable computing and fault tolerance group (TSF) and relates to safety assessment, fault removal and tolerance in safety critical autonomous systems.
Risk analysis and safety argumentation for autonomous systems
We investigate which system behavior models and human-system interaction models are appropriate for early risk analysis of autonomously-adapting systems (mostly robotics system). We particularly focus on modeling languages like UML associated with risk analysis methods.
See HAZOP-UML page.
We also investigate how uncertainties could be modeled and managed in a development process or an safety argumentation (safety case), integrating formal probabilistic models as Markov and Bayesian networks, or belief theory.
Runtime safety monitoring for autonomous systems and AI based functions
What defenses can be provided as countermeasures against the threats facing autonomous systems and Artificial Intelligent based functions? Our research on this theme focuses on the safety of autonomous systems based on the use of safety monitors.
See SMOF page (Safety Monitoring Framework)
Testing autonomous systems
How to perform tests of autonomous systems, considering the infinite input set, and that running hazardous scenarios is not possible ? We develop a method based on testing autonomous system in virtual words, and in particular a tool for generation of diverse test cases under constraints.
See TaF page (Testing Automation Framework)