Abstract

Abstract

Abstract

This paper addresses safety indicators for truck platooning at short inter-vehicle distances (with a time gap of 0.5 s). The aim of a safety indicator is to determine the correct moment for initiating a Collision Avoidance brake action to prevent a collision with the preceding truck in threatening situations. Three safety indicators are selected for an evaluation: the intended acceleration of the preceding truck, which is shared via Vehicle-to-Vehicle (V2V) communication, the Brake Threat Number (BTN — based on simple vehicle models and an emergency brake assumption of the lead), and the Time-To-Collision (TTC — based on a constant velocity assumption). The latter two do not rely on V2V communication, but are obtained via on-board signals. Requirements for the amount of false negatives (missing a threatening situation) and the false positives (identifying a safe situation as threatening) are derived from a functional safety perspective. To find thresholds for the safety indicators that minimize the false negative rate, emergency brake tests are used. To evaluate the number of false positives, a set of data of two trucks driving in a platoon at 0.5 s at mixed-traffic highways in Belgium and the Netherlands, collected during 8 hours of automated driving in a platoon, is used. The results indicate that the communicated intended acceleration of the preceding truck might be able to distinguish safe and threatening situations in a vehicle platoon. Furthermore, for situations without V2V, both the BTN and the TTC are not capable to distinguish between threatening and safe situations. The amount of false positives found in the safe driving data-set does not fulfill the requirements derived from functional safety perspective.

Abstract

rt from the optimal real-time electricity price to buy the electricity, the optimal, time dependent, capacity contracted with the DSO is of crucial importance for concerted charging of electric vehicles in a parking garage. The battery management system, on its turn, imposes constraints on the sequence of steps in which power is transmitted. Maximum power in individual charging steps has to vary as a function of the state-of-charge to keep an optimal state-of-health of the battery. Finally, the mobility wishes of the car user, given by the desired departure time and SOC will vary. In the PowerMatchingCity Smart Grid living lab [1] a strategy has been developed to optimize the charging strategies of a collection of cars by using a combination of agent-based optimization, using the PowerMatcher [2], and constrained, combinatorial optimization. In this article, this solution approach, the algorithms and the configuration are described. Furthermore, the implementation in the PowerMatchingCity [3,4] virtual power plant configuration with a fleet of 10 vehicles is discussed. First simulation results of constrained optimization for forecasting are analysed.

Abstract

This paper presents a method for trajectory control, based on feedback linearization to guide an Automatic Guided Vehicle (AGV). The novelty of this work, is the adaptation of a reference trajectory with respect to a desired velocity in real-time to avoid collisions using a time-scaling mechanism. Employing this mechanism, collision avoidance by braking is implemented. Implementation and experimental evaluation of the integrated design of trajectory control, time-scaling and collision avoidance is performed on a real AGV platform.

Abstract

Traffic light control is one of the main means of controlling road traffic. Improving traffic control is important because it can lead to higher traffic throughput and reduced traffic congestion. This chapter describes multiagent reinforcement learning techniques for automatic optimization of traffic light controllers. Such techniques are attractive because they can automatically discover efficient control strategies for complex tasks, such as traffic control, for which it is hard or impossible to compute optimal solutions directly and hard to develop hand-coded solutions. First, the general multi-agent reinforcement learning framework is described, which is used to control traffic lights in this work. In this framework, multiple local controllers (agents) are each responsible for the optimization of traffic lights around a single traffic junction, making use of locally perceived traffic state information (sensed cars on the road), a learned probabilistic model of car behavior, and a learned value function which indicates how traffic light decisions affect longterm utility, in terms of the average waiting time of cars. Next, three extensions are described which improve upon the basic framework in various ways: agents (traffic junction controllers) taking into account congestion information from neighboring agents; handling partial observability of traffic states; and coordinating the behavior of multiple agents by coordination graphs and the max-plus algorithm. © 2010 Springer-Verlag Berlin Heidelberg.

Abstract

Abstract

Abstract

Abstract