Ride the Tide of Traffic Conditions: Opportunistic Driving Improves Energy Efficiency of Timely Truck Transportation

2019 
We study the problem of minimizing fuel consumption of a heavy-duty truck traveling across national highway subject to a deadline, under a practical setting that traversing a road segment is subject to variable speed ranges due to dynamic traffic conditions. The consideration of dynamic traffic conditions not only differentiates our work from existing ones but also allows us to leverage opportunistic driving to improve fuel efficiency. The idea is for the truck to strategically wait (e.g., at highway rest areas) for benign traffic conditions, to traverse subsequent road segments at favorable speeds for saving fuel. We observe that the traffic condition and thus the speed ranges are mostly stationary within certain durations of the day, and we term them as phases where each phase is defined as a time interval with fixed speed ranges. We formulate the fuel consumption minimization problem under phased speed ranges, considering path planning, speed planning, and opportunistic driving. We prove the problem is NP-hard, and develop a dual-subgradient heuristic for instances of the scale of national highway system. We characterize conditions under which the heuristic generates an optimal solution. We carry out simulations based on real-world traces over the US highway system. The results show that our scheme saves up to 26% fuel as compared to shortest-/fastest- path baselines, of which 11% is contributed by opportunistic driving. Meanwhile, opportunistic driving also reduces driving time by 13% as compared to only optimizing path planning and speed planning. As such, opportunistic driving offers a favorable design option to simultaneously reduce fuel consumption and hours of driving. Last but not least, our results highlight a perhaps surprising observation that dynamic traffic conditions can be exploited to achieve fuel savings even larger than those under stationary traffic conditions.
    • Correction
    • Source
    • Cite
    • Save
    • Machine Reading By IdeaReader
    23
    References
    3
    Citations
    NaN
    KQI
    []