Section: Overall Objectives

Research themes

Our approach for addressing the previous challenge is to bring together probabilistic methods, planning techniques and multi-agent decision models. This will be done in cooperation with other disciplines such as psycho-sociologists for the purpose of taking into account human models. Two main research themes of robotic navigation are addressed : i) understanding complex scenes from sensors information ii) single and multi-robot planning for motion in human-populated and dynamic environments. Next, we elaborate more about these two research axes.

• Perception and Situation awareness in human-populated environment. The main problem is to understand complex dynamic scenes involving mobile objects and human beings, by exploiting prior knowledge and a stream of perceptual data coming from various sensors. Our approach for solving this problem is to develop three complementary problem domains:

  • Sensor Fusion: acquire a deep understanding on several sensor fusion problems and investigate their observability properties in the case of unknown inputs.

  • Bayesian Perception: How to take into account prior knowledge and uncertain sensory data in a dynamic context?

  • Situation awareness : How to interpret the perceived scene and to predict their likely future motion (including near future collision risk) ?

• Scaling-Up Single and Multi-Robot Motion-Planning. The challenge is to build models allowing robots to move and coordinate efficiently in dynamic environments while considering the social rules of human beings evolving and interacting with them. This requires scalable algorithms able to manage large multi-robot systems and to adapt to the dynamics. We address this problem by considering two complementary challenges :

  • Single-robot motion-planning in human-populated environment. How to plan trajectories that take into account the uncertainty of human-populated environments and that can respect the social rules of humans ? Such a challenge requires human behavior models and planning algorithms that take into account them and the dynamic of the environment (perceived following first research theme).

  • Multi-robot motion-planning in complex environments. The goal of this axis is to develop models and algorithms that provide both scalability and performance guarantees in real-world robotic systems. Our methodology builds upon complementary advantages of two orthogonal approaches, Multi-Agent Sequential Decision Making (MA-SDM) and Swarm Robotics (SR).

The Chroma project is also concerned with applications and transfer of the scientific results. Chroma have currently projects developed with industrial and start up partners. Our main application domains concern autonomous vehicles (in cooperation with Renault and Toyota), aerial robots for surveillance tasks and services robotics. These collaborations and transfers are presented later in the document.