The graphic shows work in progress to improve the quality of CycleStreets routes by taking into account the cost of making a turn.
This is part of routing enhancement work being undertaken by Codex Cambridge with us as part of the Technology Strategy Board Innovation Vouchers scheme.
Top: balanced route (amber) recommends turning right across busy Perne Road.
Bottom: by including turn cost the advice changes to straight on at the roundabout.
The upper frame shows three routes for a journey along Cherry Hinton Road, Cambridge from west to east. The quietest route in green avoids the busy roundabout by using the shared use footway and toucan crossing of Perne Road. The fastest route in red goes straight across the roundabout. The balanced route shown in amber takes the first roundabout exit and then uses the service road marked Adkin’s Corner to connect with the shared use pavement.
The problem with the balanced route is that it is rather awkward to make that right turn off Perne Road. A rider willing to make such a right turn would likely be equally confident to have gone straight on at the roundabout.
Turning right from a major road into a minor road usually requires looking behind and moving out into the road to get into the right position to make the turn. If there is a lot of traffic it may require stopping at an arms distance from the middle of the road and waiting until it is safe to complete the turn. These factors amount to an extra delay and make the route feel more busy.
The lower frame shows the situation when the cost of making a turn has been included in planning the route and shows the balanced route going straight on at the roundabout.
Choosing the numbers
In the case of the fastest route the cost of turning right can be estimated as a number of seconds. The estimate has to be for a typical rider and represent an average time. In reality the amount will vary by the time of day, amount of traffic, the type of roads involved and the rider. In CycleStreets the only data we can easily access are the types of road. So as a starting point we can create a table with the following columns:
- from road type (e.g. primary, secondary, service)
- to road type
- turn type (eg. turn left, turn right, straight on)
- estimated delay in seconds
So for instance a right turn from a primary road into a service road could have a delay of eight seconds, whereas a left turn from a residential road to another residential road might have only a one second delay.
Turn delays in seconds only affect the fastest routes, and have no affect on the quietest routes which are only concerned with how busy the route. To account for turns in quiet routes involves comparing a right turn with an equivalent cycling distance. Eg. are you prepared to cycle for an extra 100 metres to avoid a right turn off a busy road into a service road?
Choosing the right numbers for that is a bit of an art and a lot of experimentation. This is work in progress, and there’s a bit more to do before we can include this in CycleStreets routes.