Vous êtes sur la page 1sur 9

Rebalancing for Ride Sharing

Systems
ANNA MARY PHILIP
ABDHUL KHADHIR
Context
• Demands are not uniform for all OD pairs in a network
• Peak period  high bike concentration at high demand stations
• Some stations will have inventory less than the demand
• Rebalancing required to ensure demand is met at all station
• Rebalancing done in two ways
• Using trucks
• Forced redistribution by giving incentives to users
Summary of Literature
Paper Description
Chemla et al. 2013 Static truck routing problem – to find minimum cost sequence of stations to be visited.
Solved using branch and cut algorithm
Schuijbroek and Static truck based. Determining service level requirements at stations and optimal optimal truck
Hampshire 2013 routing to rebalance inventory. Solved using mixed integer programming based clustering algorithm
Erdoʇan et al. 2015 Static truck routing problem – to find minimum cost sequence of stations to be visited.
Solution algorithm based on integer programming formulation
Liu et al. 2016 Dynamic truck based. Demand prediction at stations and dynamic routing problem formulation.
Solved using mixed integer non-linear programming formulation and k-centers clustering
Chiariotti et al. 2018 Dynamic truck based. Predict requirement of bikes and docks at stations and determine most
efficient way for relocation of bikes- modelled and solved using a BDP model.
Pan et al. 2018 Static user based. Optimal pricing formulation to incentivise users, solved using reinforcement
algorithm
Haider et al. 2018 Static user and truck based. Inventory rebalancing through pricing – formulated as a bilevel problem
: agency level (minimize imbalanced stations and use truck flow there) and user level (minimum cost
problem) and solved using a heuristic approach
Physical features of the network
• Nodes representing bike stations
• Arcs representing connections between stations (bike arcs and walk
arcs)
• Demand for bikes at station i at time t - 𝑑𝑖𝑡
• Demand for docks at station i at time t - 𝑓𝑖𝑡
• Capacity of station i at time t - 𝑄𝑖𝑡
• Current inventory of station i at time t - 𝑝𝑖𝑡
• Cost of biking from station i to station j - 𝑐𝑖𝑗
Other network parameters
• Incentive for drop-off at station j instead of station i - 𝑟𝑖𝑗
𝑚𝑖𝑛 𝑚𝑎𝑥
• Assume that 𝑟𝑖,𝑗 ≤ 𝑟𝑖,𝑗 ≤ 𝑟𝑖,𝑗
• Probability of drop-off at j increases linearly from 0 to 1 as 𝑟𝑖𝑗 increases from min to max
• Incentive for pickup at station j instead of station i - 𝑟𝑗𝑖
𝑚𝑖𝑛 𝑚𝑎𝑥
• Assume that 𝑟𝑗,𝑖 ≤ 𝑟𝑗,𝑖 ≤ 𝑟𝑗,𝑖
• Probability of drop-off at j increases linearly from 0 to 1 as 𝑟𝑗,𝑖 increases from min to max
• Service satisfaction level 𝛽𝑖 at station i is penalized according to the extent of
vandalism
• Walk pickup and drop-off is possible only between neighbouring stations
1 𝑖𝑓 𝑖, 𝑗 𝑎𝑟𝑒 𝑛𝑒𝑖𝑔ℎ𝑏𝑜𝑢𝑟𝑖𝑛𝑔 𝑠𝑡𝑎𝑡𝑖𝑜𝑛𝑠
𝛿𝑖,𝑗 = ቊ
0 𝑜𝑡ℎ𝑒𝑟𝑤𝑖𝑠𝑒
Formulation
• A combination of user based and truck based rebalancing attempted
• Imbalance can be surplus or deficit
• Give incentive to users for starting a ride from a surplus station and
ending a trip at a deficit station
• Additional imbalances not met by user based rebalancing to be met
by a truck
• The rerouting (both origin and destination) can be done only for
neighbouring stations assuming the user completes the reminder of
the trip by walking from/to the neighbouring station
Objective function
The objective is to maximise the agency profit
𝑡
𝑀𝑎𝑥 ෍ ቎෍ 𝑐𝑖,𝑗 𝑥𝑖,𝑗
𝑡 (𝑖,𝑗)
Constraints
1. Capacity constraint and flow balance:
𝑚𝑖𝑛 𝑚𝑖𝑛
𝑟𝑖,𝑗 −𝑟𝑖,𝑗 𝑟𝑗,𝑖 −𝑟𝑗,𝑖
𝛽𝑖 𝑑𝑖𝑡+1 ≤ σ𝑗 𝑥𝑗,𝑖
𝑡 𝑡
− σ𝑗 𝑥𝑖,𝑗 + 𝑚𝑎𝑥 𝑚𝑖𝑛
𝑡
σ𝑗 𝑥𝑖,𝑗 − 𝑚𝑎𝑥 𝑚𝑖𝑛
𝑡
σ𝑗 𝑥𝑗,𝑖 ≤ 𝑄𝑖 −
𝑟𝑖,𝑗 −𝑟𝑖,𝑗 𝑟𝑗,𝑖 −𝑟𝑗,𝑖
𝛽𝑖 𝑓𝑖𝑡+1 for all i ∈ 𝑆
Where 𝛽𝑖 is the service satisfaction level to be provided at station I
𝛽𝑖 varies from 0 to 1, depending on the level of vandalism at i (1 means vandalism-
free)
2. Non-negativity constraint:
𝑡
𝑥𝑖,𝑗 >= 0 for all (i,j),t
References
• Chemla, D., Meunier, F., and Wolfler Calvo, R. (2013). “Bike sharing systems: Solving the static
rebalancing problem.” Discrete Optimization, Elsevier B.V., 10(2), 120–146.
• Chiariotti, F., Pielli, C., Zanella, A., and Zorzi, M. (2018). “A dynamic approach to rebalancing bike-
sharing systems.” Sensors (Switzerland), 18(2), 1–22.
• Erdoʇan, G., Battarra, M., and Wolfler Calvo, R. (2015). “An exact algorithm for the static rebalancing
problem arising in bicycle sharing systems.” European Journal of Operational Research, Elsevier Ltd.,
245(3), 667–679.
• Haider, Z., Nikolaev, A., Kang, J. E., and Kwon, C. (2018). “Inventory rebalancing through pricing in
public bike sharing systems.” European Journal of Operational Research, 1–37.
• Liu, J., Sun, L., Chen, W., and Xiong, H. (2016). “Rebalancing Bike Sharing Systems.” Proceedings of the
22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining - KDD ’16, 1005–
1014.
• Pan, L., Cai, Q., Fang, Z., Tang, P., and Huang, L. (2018). “Rebalancing Dockless Bike Sharing Systems.” 1.
• Schuijbroek, J., and Hampshire, R. (2013). “Inventory Rebalancing and Vehicle Routing in Bike Sharing
Systems Inventory Rebalancing and Vehicle Routing in Bike Sharing Systems.” (Cmmi), 1–27.

Vous aimerez peut-être aussi