Safety and Traffic Study of Automated Shuttle Interactions in City Traffic

Presenter

Étienne Beauchamp has his Bachelor in Electrical Engineering from École Polytechnique de Montréal, Étienne is currently a master's student in Civil Engineering under the supervision of Professor Nicolas Saunier.  Specializing in the use of embedded systems and data valorization, his work focuses on intelligent transportation and its impact on road safety.

Title of Abstract

Safety and Traffic Study of Automated Shuttle Interactions in City Traffic

Background

Driving automation is happening at a rapid pace, whith different driver assistance systems already available in mass market cars. Furthermore, the interest amongst the research community is high, as a class 4 autonomous taxi service (according to the J3016 standard) is already offered in the United States and autonomous shuttle projects are now present around the globe. However, this rapid development in driving automation leads to concerns and questions about their impact on traffic and safety, in particular for vulnerable road users. In these rapidly changing conditions, it is essential to be able to quickly assess safety, which can only be done with observations or surrogate measures. These indicators also help to better understand the factors that may lead to collisions. The Québec Road Safety Code was amended in 2018 to include the definition of an autonomous vehicle and to allow the development of pilot projects aimed at carrying out experiments involving autonomous shuttles. Vehicles of this type were notably tested in 2019 in Montreal and Candiac (on the South Shore). These two locations therefore provided an important opportunity to study these vehicles and their interaction in real traffic.

Aims

The proposed research project aims to study the interactions of road users, motorized and non-motorized, with an autonomous shuttle, in terms of flow and safety. This breaks into two specific objectives:
1. to analyze the safety of the interactions identified, through surrogate measures of safety, mainly for vulnerable road users;
2. to assess, with different indicators, the effects on traffic flow.

Methods

Since road safety studies are usually based on the analysis of public roads accident data, it is difficult to obtain a sufficient amount of data in short periods of time. In addition, analyzing accident data involves waiting for accidents to happen before one can draw conclusions and implement counter measures. Because of its novelty, this study is based on the analysis of traffic events which do not involve a collision like interactions and conflicts from field observational data. Cameras have been installed at various locations on the Montreal and Candiac shuttle routes. Data is extracted from the collected videos through computer vision and machine learning techniques. This data is in the form of road user trajectories, i.e. their position and speed at each instant. Various traffic and safety indicators are then derived and analyzed using statistical tools.

Results

While the autonomous shuttle project in Candiac has been considerably delayed and data collection is currently ongoing, few results for this site are currently available. The study is currently at the stage of data cleaning. This data was extracted for the Montreal sites and preliminary results have been obtained. At present, it is estimated that the shuttles could have a regularizing effect on the road and lead to better road safety, in particular for vulnerable users. However, it is expected that the very conservative behaviour of these shuttles may induce more aggressive behaviour among certain users. Still, these remain hypothesis that need to be explored and validated from the whole dataset.

Discussion

The field study and the initiation of the literature review led to the following observations. First, this kind of shuttle present some problems with the road and weather conditions in Québec. The two manufacturers of the vehicles in question are located in France, which notably led to significant delays in the event of technical difficulties. This was one of the main causes for the significant delay faced by the project in Candiac. Also, weather conditions and certain changes on the circuit can lead to unexpected difficulties. This raises questions about the transferability of automated technology and the feasibility of fully automatic service.

Conclusions

For the moment, the research project is ongoing. Complete results and conclusions will be presented at the conference.