In any road transportation system, the intersection and its controls represent the locus of greatest complexity and risk. This will also be true for public-area mobile robots (PMRs) that operate in the public right of way.
At traffic intersections, a PMR might be situated at the road shoulder (curb/kerb), on the sidewalk (including the ramp), in the crosswalk, or within a bike lane. In all cases, we require an agreed set of standard behaviours that may be executed according to onboard navigational software or a body of remote oversight or teleoperator standards, or a combination of these.
In 2024 PMR navigational behaviours are managed using control systems ranging from 100% teleoperation to a very high degree of intelligent software mediation. (Some delivery robot providers claim well over 90% autonomy but their evidence is proprietary). In any case, there are no civilian, mobile, ground-capable devices which are autonomous in the sense of operating exclusive of human oversight or intervention.
Regardless of the locus of navigational control, we still expect a body of understandable, reliable, and regulated device behaviours. For the remainder of this blog post, we are not concerned whether a device is controlled by software, a teleoperator, or both. Rather, we expect a standard body of behaviours to be exhibited. These standard behaviours would impact:
Traffic flow, including crowd and congestion control;
Safety, including safety for the device as well as proximate bystanders and vehicles;
Security, including what such a device might do if commandeered or what it might carry if misused; and
Monetization, including methods to identify and ensure payment of licensing and user fees.
What Can We Learn from Cars and Trucks?
Motor vehicular traffic is managed by a combination of five elements:
roadway configuration
traffic signals and signs
traffic rules (motor code)
driver training
enforcement
Each of these five elements has or requires an analog for the management of PMRs.
Pathway Configuration
For PMRs, roadway configuration is replaced by pathway configuration comprised of one or more of walkway, sidewalk, crosswalk, bike lane, or road shoulder which, when taken together, is generally a more complex and variable machine-mobility environment than is a roadway for motor vehicles.
Traffic Signals and Signs
For PMRs, there are relatively fewer traffic signals and signs when in each of walkway, bikeway, or roadway. A PMR is expected to use the signals and signs and follow the rules for pedestrians, cyclists, or road vehicles, respectively.
Traffic Rules
Many of the rules that apply to the walkway, including critical micro-behaviours at intersections and within crosswalks are social such as maintaining shy-distances, moving aside, and granting right-of-way. For PMRs, these are expected to be managed by software or decisions made by teleoperators and carefully controlled.
Training
For PMRs, driver training is replaced by intelligent software and teleoperator training. This has two problems. There is no current process to independently certify that PMR navigation and behavioural control software operates reliably and according to local traffic and social rules. Second, teleoperators are human and subject to the same attention shortcomings and self-competitive motivations that afflict human drivers. As an example, it has been reported that at least one delivery fleet operator incentivizes its teleoperators for delivery speed, which has given rise to behaviours wherein PMRs have rushed intersection crosswalks as a traffic signal turns amber.
Enforcement
PMRs rely on various, non-standardized combinations of pre-recorded digital maps and remote operation via cameras and other sensors; their navigational behaviours are expected to be constrained by geofenced maps, onboard planning software, and remote operation skills. This provides a control environment from which we may expect a high degree of conformity—certainly higher than we currently experience from motor vehicles operated by independent, competitive humans that may operate under sub-optimal cognitive and attention conditions.
Nearly everything about PMRs can be managed at the traffic intersection
Travelling along a linear pathway that is structurally suitable for PMR mobility is not a major challenge and is not our principal concern. The majority of the complexity and risk occur at roadway intersections. We need to be especially certain of pedestrian and cyclist safety there, as road crossings involve sharing access with full-sized motor vehicles many of which are competing to cross the intersection, and many of those are making risky turning manoeuvres.
In terms of traffic flow, safety, security and monetization, intersections provide network nodes for detection and observation. The potential for large and various robot fleets from many independent operators raise the following issues:
Crowd Control
It will be necessary to limit the number of PMRs at an intersection, crosswalk, or curb ramp. No individual robot operator can do this without coordination among all other traffic. This will require a regional ground traffic control system analogous to air traffic control. One is proposed in ISO draft technical standard DTS 4448-5. However, this will not be required until multiple PMR operators send PMRs to the same intersection at the same time.
Liability
In the event of a crash or other event, it is not possible to guarantee that the PMR involved will have recorded or be able to record all aspects of an event. This is important because there is no proximate human to speak for that robot. It is not clear that we can rely on these devices or their fleet operators to ensure that all appropriate information is captured and disclosed. Several parts of ISO DTS 4448 address aspects of this, but these drafts are currently unpublished.
Device Registration
It is critical that PMRs, whether licensed to a commercial enterprise, privately owned, or part of a municipal public works fleet, be identifiable and that no unregistered PMRs be permitted to operate freely in public space. A proposal for this is included in ISO DTS 4448–2 but is currently incomplete.
Security
This encompasses physical, operational, and cyber security. Several parts of ISO DTS 4448 address various aspects, but these are still in draft form.
Preparing for Safe Deployment of PMRs
Cities that wish to be prepared for the safe deployment of PMRs will need to focus on intersection management. System components to support these requirements will need to be installed at intersections to capture information such as IDs, speeds, positions, and behaviours. Such sensors and recorders will be integrated to systems for crowd control, behaviour checks, liability, confirmation of device registration, hazardous goods registration, and crash recording, as well as systems for enforcement, payment collection, and others.
One critical purpose of ISO DTS 4448 is to provide standardized procedures and metrics regarding the behavioural outcomes for the management systems this equipment would support and the PMR fleets it manages.
Regarding the drafting of all remaining parts of ISO DTS 4448 that address these and many other aspects, we invite interested parties to become members of the Urban Robotics Foundation to review and contribute to shaping the final version.