A Look-ahead Car Following Scheme for Efficient Driving on Urban Roads

摘要

For safe and efficient driving of a vehicle on urban roads, it is essential to analyze the trends of the vehicles ahead to take early measures in changing traffic situations. Existing efficient driving systems based on optimal car following compute the vehicle control input by solving an optimization problem over a prediction horizon, and at the expense of large computation cost, they provide significant improvement in traffic flows and fuel consumption. This paper proposes a look-ahead car following scheme, which can take anticipatory driving decisions with negligible computation cost, for efficient driving of a vehicle. Specifically, at first, the distinctive features of an optimal car following scheme over traditional car following of a human driver are investigated. Then, based on the features observed, a look-ahead car following scheme is formulated that can partly reflect the desired driving characteristics of the optimal car following scheme. The proposed scheme extends a traditional car following model by incorporating the predicted state of the preceding vehicle in a restricted look-ahead horizon. Finally, the proposed look-ahead car following scheme is evaluated in typical urban traffic scenarios, and the observed driving characteristics and performances are compared.