Portwise’s view on autonomous vehicle modelling: Q&A with Arjen de Waal
Autonomous vehicles offer exciting potential for terminal logistics, but their success depends on far more than simply navigating from A to B. In complex port environments, these systems must operate safely and efficiently among other equipment, under real-world constraints. In this Q&A with Arjen de Waal, Head of Simulations at Portwise, we explore the key challenges and lessons learned in simulating autonomous vehicles at ports; from mixed traffic modelling to early insights on layout and control.
As autonomous vehicles will find their own way from A to B, what makes it relevant to model such vehicles in port operations?
To be effective in a port operation, autonomous vehicles need to do their jobs in time and in the correct sequence. Also, the vehicles should avoid causing hindrance of other, manned and unmanned equipment doing their jobs in the same port environment. In other words, there is more to it than “just” driving from A to B and not hitting anything while “en route”. The effectiveness of autonomous vehicles in port operations are affected by port layout, systems linked to autonomous vehicles, operational procedures, traffic rules, etc., all of which can be explored well in a simulation model.
As the development of autonomous technology is still ongoing and actual implementations of such technology are still rare, how to account for the unknowns in the modelling?
Portwise has defined certain fundamentals that we believe should apply for operating autonomous vehicles in reality in mixed traffic in port environments. The same fundamentals are used in Portwise’s simulation models. Those fundamentals can be validated with vendors of autonomous technology and tuned to specific situations. With this basis, the simulation model serves as a great starting point to investigate those unknowns in an early stage in a safe and inexpensive lab environment.
When modelling fleets of autonomous vehicles, what are major differences compared with centrally orchestrated AGVs?
An obvious difference is that autonomous vehicles are expected to operate in “mixed traffic” of manned equipment whereas AGVs are expected to operate in a fenced off area where no manned equipment is allowed. The autonomous vehicles use sensor technologies to find their way without hitting any object whereas a central system, the ECS, manages for a fleet of AGVS all individual AGV routes. This makes for a different modelling approach for the autonomous vehicles compared to the AGVs. One example: as the autonomous vehicles operate on the same infrastructure as manned vehicles, it is important that the autonomous vehicles obey the same traffic rules that apply to manned vehicles. This is obviously not relevant for a centrally orchestrated AGV system.
What are some of the lessons learned so far in modelling autonomous vehicles?
One of the lessons learned is that different driving directions in a port environment could lead to less crossings of vehicles. This leads to less hindrance of vehicles, smoother overall traffic flow and higher performance, even if it may also lead to longer vehicle driving distances. We also found that it is important that there are clear priority situations that both manned and autonomous vehicles must know, understand, and obey. Also, autonomous vehicles should predict whether they are able to cross certain areas completely with no risk of getting stuck in the middle of a busy intersection, for instance. All these rules we found to be important are included in our simulation models.
What is the most promising port set up for autonomous vehicles?
This is too early to tell as there are still many things unexplored. Also, the autonomous technology to support larger fleets of autonomous vehicles in mixed traffic port environments is still being developed and improved. Furthermore, lessons learned of the deployment of this technology in reality are still scarce. In the end, we like to feed these lessons learned back into our models for validation purposes for future applications. As such, we can actually support the development of this technology for effective use in port operations.
About the Author: Arjen de Waal, Head of Simulations at Portwise
Arjen de Waal leads the Simulation Department at Portwise. With over a decade of experience, he has supported terminal projects worldwide and is the driving force behind the Portwise simulation library. Arjen holds an MSc in Operations Research from the University of Amsterdam and co-develops the Portwise Academy, the firm’s internal training programme.