Work With the Street
In dense neighbourhoods, vehicles spend up to 68% of shift time stationary at the kerb while the courier walks corridors, climbs stairs, or waits for intercoms. Route optimisation must solve for legal stopping points, not just road speed.
Why Standard VRP Solvers Fail Inside City Centres
Classical Vehicle Routing Problem (VRP) algorithms minimize driving time between latitude/longitude pins. In suburban sprawl with private driveways, that assumption holds. In dense districts like Islington, Amsterdam-Centrum, or Milan's Area C, a latitude/longitude pin on a building roof tells you nothing about where a 3.5-tonne van can legally halt without incurring a £65 Penalty Charge Notice (PCN) or blocking a bus lane.
Modern urban routing engines separate the Vehicle Parking Node (Kerb Pin) from the Final Doorstep Node (Vertical Pin). One well-chosen loading bay stop often serves seven distinct commercial and residential doorways within a 90-metre walking radius—turning a congested driving loop into a structured "park-and-loop" cluster.
Modal Filters, ANPR Bollards, and Permeable Routing
Across European and UK cities, Low Traffic Neighbourhoods (LTNs) and circulation plans (such as Ghent's Circulation Plan or Barcelona's Superblocks) intentionally sever through-traffic for motor vehicles using planter boxes and Automatic Number Plate Recognition (ANPR) cameras, while maintaining full permeability for cycle logistics. This creates two distinct routing graphs over the exact same geography:
- The Motorised Graph: Must enter and exit individual neighbourhood cells via boundary arterial roads, turning a 200-metre crow-flies distance between two parallel streets into a 1.9-kilometre detour through three signalised junctions.
- The Cycle-Freight Graph: Passes straight through modal filter gaps (typically 1.2m to 1.5m wide), maintaining a steady 14 km/h door-to-door average speed and parking directly on broad footways or dedicated micro-loading bays without hunting for kerb space.
Accounting for "Vertical Last Mile" Time
Two stops with identical parcel counts can vary by 400% in service duration depending on building morphology. A street-level terraced house with a letterbox takes 35 seconds. A 14th-floor flat in a security-gated tower block with a slow service lift and concierge sign-in book takes 6.5 minutes. Systems-minded planners attach a Building Friction Coefficient to every postcode or UPRN (Unique Property Reference Number) so route wave planners do not overload couriers serving high-rise estates.
| Neighbourhood Morphology | Optimal Vehicle Mode | Avg Stops / Hour | Kerb Dwell Share | Key Routing Rule |
|---|---|---|---|---|
| Historic Core / LTN Cell | E-Cargo Trike (250 kg payload) | 24–32 stops/hr | 42% | Route through modal filters; reload via swap-pods |
| High-Rise Residential Estate | Compact EV Van + Sack Barrow | 14–18 stops/hr | 76% | Group by tower core; consolidate parcels into tote bags |
| Timed Commercial High Street | 3.5t Box Van (Pre-10:00 Window) | 11–15 drops/hr | 68% | Hard time-window constraints before pedestrian curfew |
| Outer Victorian Terrace Grid | Light EV Van + Cargo Bike Split | 22–28 stops/hr | 55% | Send oversized (>10 kg) on van; small packets on bike |
Dynamic Micro-Reloading vs. Returning to Base
Cargo bikes excel in stop density but are constrained by cube (1.5 to 2.2 cubic metres) and payload (150 to 300 kg). Rather than forcing riders to pedal 2.5 km back to the depot mid-shift, operators deploy pre-sorted swap-containers to a quiet kerbside rendezvous point or partner micro-locker. Every handoff requires strict scan custody so parcels never disappear between vehicles—explained next in Keep the Local Picture Current.