Driving behaviour improvement using Mobileye CAS
A 13 weeks test was conducted to understand how Mobileye, data collection and driver engagement can improve driving behaviour.
Prime movers, Dump trucks
15 units
To understand and show that collision avoidance system such as Mobileye and human factor can co-exist to deliver improvement in driving behaviour with continuous engagement. 15 drivers were randomly selected in partnership with 2 companies in container logistics and construction. For the objective of this test, no incentives or penalties were introduced to the drivers.
The test was conducted on a set of 15 drivers involving a mix of prime movers and dump trucks over a period of 13 weeks clocking 200,000km. Mobileye units were initially fitted on these vehicles with its audio and visual warning disabled. In this ‘silent’ phase, the drivers were not given any warning. The drivers’ base behaviour were then logged for each silent Mobileye warning trigger that was sent through a vehicle GPS telematics device for data capture in Acudrive platform.
After 4 weeks, the Mobileye units were switched on to ‘live’ mode, providing audio and visual warning to the drivers. At the same time, a briefing of the various Mobileye warning type was conducted for the drivers. Approximately 2 weeks into the live phase, a sharing session was held with the drivers to go through the data captured and how some of the undesired driving behaviours can be improved. This live phase continued for 8 weeks, logging the driver’s behaviour with relation to the Mobileye warning, like before.
Mobileye Events | Silent phase | Live phase | Observations |
---|---|---|---|
Headway Warning | 20.61 | 15.81 | 23% Reduction |
Forward Collision Warning (Speed > 30km/h) | 0.42 | 0.21 | 50% Reduction |
Lane Departure Warning | 53.08 | 18.87 | 64% Reduction |
Pedestrian Collision Warning | 0.03 | 0.02 | 33% Reduction |
Urban Forward Collision Warning (Speed < 30km/h) | 2.21 | 1.89 | 14% Reduction |
The table above outlines the the events captured (average for 15 vehicles, and normalized to per 100 KM traveled) for both silent mode and live mode, based on best weekly scores in both phases. It is encouraging to see a behavioral improvement for both heading warning (ie. following too close) and lane departure warning (ie. changing lane without giving turn signal). These two unsafe habits are the main reasons for rear-end and side swipe collision on the road.
From Individual driver perspective, there was also reduction between 14% to 78% in overall warning count. This disparity is due to each driver’s different base behaviour and clearly shows the upside from a high base count. The test allowed the identification of different driver’s behaviour in operating the same type of vehicle and operations. With this information, individual driver behaviour improvement can be better targeted and managed. Such insights support the introduction of a driver risk assessment framework to mitigate behaviour risk while on the road, including any safety incentive program.