Route Optimization
Route optimization is the task of computing the best routes for a fleet of vehicles under a given set of constraints.
The goal is not only to shorten distance: vehicle count, time each passenger spends on board, walking distance to the stop, shift times and vehicle capacity are all weighed at once. A human planner cannot hold all of these at the same time, so improving one usually breaks another. Optimization settles that balance by calculation.
What problem does it solve?
In a shuttle operation the questions are nested inside one another. How many vehicles go out? Which passenger boards which vehicle? Where do the stops go? In what order are they served? Every answer changes the others. Moving a stop 200 metres can push a passenger onto a different vehicle, and that vehicle may then exceed capacity.
The classic textbook routing problems address only one of these questions, in a simplified form: a single vehicle, fixed capacity, an unchanging road network, known and constant travel times. Field conditions are not like that. A shuttle plan has to survive a passenger list that changes weekly, traffic that varies by the hour and vehicles swapped at short notice.
So the optimization that works in practice is not the one that solves a single problem perfectly, but the one that weighs these questions together against real data.
- Assigning passengers to stops and placing the stops themselves
- Determining vehicle count and each vehicle's passenger group
- Deriving the stop visit order
- Satisfying capacity, maximum ride time and shift times simultaneously
Route optimization for employee transport
In employee transport, departure and arrival times are set by the shift and are not negotiable. Passengers and their addresses are known. The real question is how many vehicles and which stops will get everyone to their shift on time.
Optimization assigns each employee to the nearest stop within the walking-distance limit, merges nearby stops and builds the fixed route with the fewest vehicles. Fewer stops shortens ride time but lengthens the walk; when those two constraints are not balanced together, the plan either runs late or generates complaints.
Route optimization for school transport
School transport changes the picture: students are collected at the door, merging stops is usually not an option, and the maximum ride time is far more restrictive than in employee transport. Parent tolerance is low.
Here optimization builds door-to-door routes, works backwards from the school start time and keeps every student's time on board within the limit. Where students from different schools in the same area can share vehicles, the number of vehicles on the road drops markedly.
Why straight-line distance misleads
Straight-line distance between two points is easy to compute and wrong. Vehicles do not travel in straight lines: one-way streets, divided roads, turn restrictions, bridges and junctions all lengthen the real distance. Within a city the real road distance runs markedly above the straight line, and the gap is not the same ratio on every route.
Proper optimization computes on the real road network, with turn restrictions and historical traffic data that varies by time of day. A travel time valid at 07:30 is not valid at 10:00; when a plan is built around a shift time, the travel times it uses must belong to that hour.
The vitaRoute approach
vitaRoute's routing engine is a reinforcement-learning based service routing model developed by vitaRoute within a TÜBİTAK-supported R&D project. It weighs constraints together rather than in sequence, and learns from every optimization on the platform, every manual correction and every trip actually performed.
In practice this means the plan is fed by your operation's own history rather than a fixed formula. Two customers in the same city will not get the same plan, even under the same constraints.
- Walking distance and gradient limits — 300 metres uphill is not 300 metres on the flat
- Mixed fleet — vehicles of different capacities used together in one plan
- Locked routes — the routes you do not want changed are preserved exactly
- Zone rules — constraints defined for particular districts or campuses
- Real road network and time-of-day historical traffic data
Measurable results
The figures below come from vitaRoute operations. They show the difference obtained with the same passenger count and the same stops, changing only the plan.
- −93%
- Planning time
- −32%
- Fuel cost
- %87
- Vehicle occupancy
- maks. 10 sn
- Computation time
3 hours → 12 minutes
at the same passenger count
54% → 87%
for 1,000 passengers
Frequently asked questions
Are route optimization and route planning the same thing?
Route planning is the whole process of producing a plan; route optimization is the computational part of it. Planning can be done by hand, whereas optimization requires a calculation that weighs the constraints together.
At what passenger count does optimization become necessary?
With a single vehicle and a fixed passenger list, planning by hand may be enough. As soon as there are several vehicles, a changing passenger list or a variety of shifts, a hand-built plan falls behind quickly, and from that point the difference becomes measurable.
Do we have to replace our existing GPS system?
No. vitaRoute connects to your existing GPS provider over a read-only connection; no additional hardware is fitted in the vehicle. Raw position traces are not stored — only the trip verification result is kept.
How long does a plan take to produce?
At a scale of 1,000 passengers the computation takes at most 10 seconds. The hand-built plan it replaces takes hours.
Are my manual corrections lost?
No. Locked routes are preserved exactly, and manual corrections become part of the data the engine learns from, so they are taken into account in later plans.
Try it with your own data
Upload your employee or student list, define your stops and shift times, and see the first plan within minutes.