Overview

The interACT project worked on the safe integration of automated vehicles into mixed traffic environments by designing, implementing and evaluating solutions for safe, cooperative and expectation-conforming interaction of the Automated Vehicles with both its on-board user and other road users. With a total budget of 5.5 million Euros funding by the European Commission, 8 partners from 4 European Countries worked together from 2017 to 2020 and joined their expertise to contribute to the vision of designing a cooperative interaction of automated vehicles with other road users in mixed traffic environments.

As the interACT project was coming to its end, our consortium had organized its Final Event on 1st April 2020 to be held in Munich Germany. In light of COVID-19 breakout and related containment measures, the interACT consortium decided to turn the physical interACT Final Event into a Virtual one as to secure all participants’ safety and health.

Within this context, the consortium organized a two-day online event on the 18th and 19th of June at 13:00 CEST in order to present its produced results after more than 3 years of research work made. During the online event, the interACT project partners presented the most important interACT scenarios; the acquired knowledge after observing human interaction in real traffic; the external eHMI of two test vehicles of the project for the safe integration of automated vehicles in mixed traffic and the results from the impact assessment and impact studies in the areas of road safety and traffic.

Key Results in Brief

Results Human interaction behavior

  1. Definition of interaction terminology
  2. Several observation studies on human-human interaction in Greece, Germany and the UK
  3. Traffic participants tend to avoid conflicts; Interactions are more likely to occur when the vehicle is driving slowly; Pedestrians mostly focus on implicit vehicle cues rather than explicit communication

Results Intention recognition

  1. Risk analysis framework for the prediction of traffic participants location
  2. Pedestrian intention prediction using the semantic map and behaviour models of other traffic participants
  3. Novel deep learning techniques, for classification of pedestrians’ head orientation and hand waving gestures
  4. Hidden Markov model for vehicle maneuvers recognition and generation of intention-aware trajectory.
  5. Extended vehicle prediction trajectory via fusion of intention-based with typical motion-based

Results Communication and Cooperation Planning Unit

  1. Recognition of traffic conflicts between Automated Vehicles and other traffic participants
  2. Implementation of reaction strategies according to the identified situation (future path constraints, candidate actors for HMI/eHMI interaction)
  3. Integration of internal and external HMI to enable human-like interaction
  4. Development of safety layer for emergency situations

Results HMI/eHMI

  1. Two interaction strategies defined: intention-based & perception-based strategy for HMI/eHMI
  2. Two eHMI technologies developed and implemented: 360° Light Band & Directed Signal Lamp
  3. Two iHMI technologies: Light Band & Automation Display

Results Evaluation methodologies

  1. Evaluation criteria and methodologies derived for Automated vehicles
  2. interACT demonstrators evaluated in test-track studies, while eHMI/iHMI solutions were also evaluated using driving and pedestrian simulator
  3. Impact assessment carried out to understand the effects of the interACT solutions on safety, traffic flow, criticality, comfort, and acceptance