Skip to Main content Skip to Navigation

Governing Automated Vehicle Behavior

Abstract : From a societal point of view, autonomous vehicles (AV) can bring many benefits, such as an increase of road safety and a decrease of congestion. However, due to the highly dynamic environments these vehicles an encounter, technical concerns about their feasibility and ethical concerns about their actions’ deliberation, explainability and consequences seems to damper the enthusiasm and optimism that once ruled the automotive sector and the public opinion. This thesis addresses the core questions surrounding autonomous vehicles today: how they should account the road users sharing the same environment in its decision making and also how they should deliberate to act in dilemma situations. Firstly, a decision-making process for the autonomous vehicle under general situations is proposed, and then, as a particular situation, the deliberation under certain collision, with other vehicles, pedestrians or static objects, is treated. Then, to relax the assumption that the behavior of the other road users does not change during the execution of a policy, used to implement the decision making, the intent of each one of them are estimated comparing the prediction made by a Kalman filter and the real-time observation. To account the interaction between road users an incomplete game model is proposed, using the intent probabilities calculated beforehand, finally producing a coherent estimate of the evolution of the other road users.
Complete list of metadata
Contributor : ABES STAR :  Contact
Submitted on : Thursday, June 30, 2022 - 1:02:02 AM
Last modification on : Wednesday, August 3, 2022 - 3:59:04 AM


Version validated by the jury (STAR)


  • HAL Id : tel-03709568, version 1


Nelson De Moura Martins Gomes. Governing Automated Vehicle Behavior. Computers and Society [cs.CY]. Sorbonne Université, 2021. English. ⟨NNT : 2021SORUS169⟩. ⟨tel-03709568⟩



Record views


Files downloads