Context Modelling applied to the Intelligent Vehicle Navigation

Abstract

This paper faces the problem of intelligent vehicles in interaction with their occupants and the environment, by modelling the semantic context associated to the navigation. With a semantically modelled context, an intelligent vehicle will not only drive itself safely, but it will also be able to reason on the situation and act accordingly. To do so, it is necessary to first define the Context of Navigation, and then to set the inference rules for it, in order to enrich the robot's comprehension of the situation. In this paper we propose our definition of the Context of Navigation, based on the information that could be needed by the vehicle's controller. We split it into two components: the Dynamic Context and the Static Context. In this paper we will focus on the latter. We then model the Static Context of Navigation of the autonomous driving -for instance the passengers' driving preferences -and to make the robotic car adapt its behaviour to this new information in real time. Finally, a short practical example of our proposition is shown and discussed.

Autonomous vehicle navigation with semantical modelling: A project overview

“…We presented a methodology to model the context using Ontologies and a reasoner basis model [19]. This context structure was, then, used in association to an autonomous vehicle navigation methodology, based on the IDWA method [1], generating a new robotic vehicle navigation methodology called ICDWA [20]. Validation results proved the pertinence of our method, leading us going forward if our researches in how to making the autonomous vehicle "sensible" to the context involving the navigation task.…”

“…2) Directly integrating the reasoner recommendations in the IDWA, as indicated in Fig. 5, is an improvement of the previous blending control rule, which considers a new proposal of the IDWA in order to take into account the context recommendation, as we proposed in [20] With 𝛿 ∈ ℝ. We called this new method, ICDWA (Imageand-Context-based Dynamic Window Approach) [20].…”

“…5, is an improvement of the previous blending control rule, which considers a new proposal of the IDWA in order to take into account the context recommendation, as we proposed in [20] With 𝛿 ∈ ℝ. We called this new method, ICDWA (Imageand-Context-based Dynamic Window Approach) [20]. This new approach considers the longitudinal velocity "v" and the yaw velocity "w" as control navigation inputs to the robotic vehicle, so, in this case, the context reasoner recommends (Vcn,Wcn) as context navigation inputs.…”

“…The weight 𝜀 ∈ [0,1] indicates the importance of the context recommendation in the total optimization procedure. Let's consider in this explanation only the longitudinal velocity criteria, the formulation to the yaw velocity is similar, and considering context(v) defined in terms of the sigmoid function, We review here the results presented in [19] and in [20] in order to show the application of the context modelling and reasoning in the autonomous navigation of a robotic vehicle. We have onboarded our IDWA navigation control method, as described in section III-A in the professional driving simulator SCANeR Studio, Fig.…”

“…The second validation test was performed using the second methodology recalled in section III.C.2. In this case, the autonomous navigation loop onboarded in SCANeR Studio was the new methodology ICWDA proposed in [20]. The navigation scenario consists of an autonomous avoidance of static obstacles in the lane, with a lane change, as shown in It is shown the decelerations when the obstacles are detected and accelerations during the overtaking maneuvers.…”