Is delivery by driverless vehicles soon to be a reality? While already being used for logistics purposes in limited areas on certain sites, the use of autonomous vehicles for delivery in open spaces poses several challenges to its immediate deployment.
While autonomous vehicles are already a reality on sites closed to the public, using them for delivery purposes in open spaces requires hugely complex systems. In 2014, SAE International defined the six levels of autonomous driving, ranging from traditional driving with zero automation (level 0) to total autonomy (level 5). It is the latter that is commonly being referred to when talking about autonomous driving. Already in use where local regulations allow (such as in the USA, where the start-up Nuro operates, having launched the first autonomous delivery vehicle in 2018), the main issue remains liability in the event of an accident causing personal injury when there is no driver. On top of this is the cyber risk, since these vehicles rely heavily on computer systems.
Major safety issues for delivery
Where they are currently being used (mines, quarries, inside warehouses, and on container port terminals), autonomous vehicles are capable of providing a continuous service, with no need for human operators.
In all these environments, the autonomous vehicle operates in a delimited area where the intervention of other users is controlled and limited to what is strictly necessary. Contrary to this, for urban distribution the autonomous vehicle must take into consideration multiple random actions by other users, who are often vulnerable and not necessarily respectful of road signs. In addition to posing a safety issue, this represents a real development challenge, i.e. being able to authorise the presence of these vehicles on the road without endangering other users of the public space.
Autonomous vehicles that adapt to the city
In its current state, an autonomous vehicle must call on a wide range of equipment to be able to move around. A GNSS (GPS, Galileo, etc.) helps it to locate itself, but it can be ‘blinded’ by the environment. This is why constant scanning of the vehicle’s surroundings is essential, so it may detect obstacles and understand the area in which it is driving. It uses lidar and radar systems as well as cameras to do this. Information from these devices is then processed by software in order to adapt the vehicle’s autonomous driving while ensuring everyone’s safety. 95% of accidents are caused by human actions, thus justifying the use of electronic safety systems. However, although immune to distractions, electronics and IT are not perfect. Coding error or the obstruction of a sensor is always possible.
Despite these major issues in the autonomous vehicle’s development, the prospect of last mile delivery and the challenges of urban logistics are spurring on the search for solutions. In 2018, the German manufacturer ZF revealed its autonomous urban delivery van. Like a dog on a lead, the van follows the postman as he delivers parcels to two neighbouring addresses. It is also able to find a parking space and park in it without assistance. The technology is therefore able to provide the necessary autonomous driving solutions for city deliveries.
Although the autonomous delivery vehicle will not necessarily be ready for use tomorrow, encouraging results are emerging all over the world. In 2019, the American retail giant Walmart announced it was starting its first tests of autonomous delivery vehicles. In 2021, a new phase of testing will start in Arizona.