There is currently a high desire by manufacturers to introduce Automated Vehicles (AVs)to the market. As AVs are likely to be deployed in mixed traffic, they need to interact safely and efficiently with other road users, including other vehicles, cyclists and pedestrians...
There is currently a high desire by manufacturers to introduce Automated Vehicles (AVs)to the market. As AVs are likely to be deployed in mixed traffic, they need to interact safely and efficiently with other road users, including other vehicles, cyclists and pedestrians. Although obstacle detection by AVs is almost flawless, the AVs cannot communicate their intentions to other road users. This limitation currently reduces their appeal and value to the user. To ensure intuitive and cooperative interaction between AVs and others, and a smooth flow of all traffic, it is essential that there is good means of communication between all actors – the AV, the on-board user and the other road users. The main objective of interACT is to substantially improve this cooperation strategy of AVs.
For achieving this, the interACT project works on the following objectives:
Develop social-psychological models to compile a catalogue of interactions, identifying the main communication needs of road users in current and future traffic
Improve software algorithms and sensor capabilities for enhancing intention recognition and behaviour prediction of other road users
Develop a Cooperation and Communication Planning Unit to integrate planning algorithms, providing synchronized and integrated communication protocols
Ensure safety of road users by developing easy-to-verify software for a safety layer, and novel methods for fail-safe trajectory planning
Establish and refine evaluation methods for studying interaction of road user with AVs and user acceptance and assessing cooperation.
interACT develops prototypes that will be demonstrated and evaluated in simulators and two test vehicles, assessing ease-of-use, acceptance, safety and reliability. The impact of interACT solutions is being investigated and these are being exploited to the AV market by the industrial partners.
The project started by defining its use cases and scenarios for which the interaction between AVs, the on-board user and other road users have a high impact. This led to four urban must-have use cases related to intersections and parking areas. Thereafter, we firstly conducted our observational studies on current human-human interaction at three sites in the UK, Greece and Germany. Human interaction behavior was assessed to better understand what the key characteristics of behavior in mixed traffic environments are. Thedata were used to identify the interaction requirements and thresholds, to model these interactions and to enhance intention recognition algorithms. Secondly, we started to develop the Cooperation and Communication Planning Unit (CCPU). This is the central intelligence for enabling the vehicle to interact with its on-board user and the other road users according to their expectations. The CCPU consists of four software modules and one enabler: Situation Matching, Interaction Planning, Trajectory Planning and Safety Layer modules as well as the Scenarios’ digital catalogue. The software components are available in a first working version and will be integrated in the demonstrator vehicle.
In parallel, we worked on the Human-Machine Interaction (HMI) solutions to govern the interaction of the AV with others. Based on the outcomes of the observational studies we derived generic interaction strategies and general HMI messages to enhance the cooperation and safe interaction of the AV. We further started to develop technical HMI hardware solutions.
For the upcoming project months we are intensifying the work on the integration of the software and hardware components into the vehicles and the preparation of the evaluation studies. The demonstrator vehicles by CRF and BMW will show different results including a fully functional CCPU and fully integrated external HMI components and will be evaluated on test tracks and in a real world study.
For dissemination and exploitation several activities have been undertaken. interACT has a well-perceived presence on the internet with a website, linkedin , twitter, facebook group and has distributed two e-newsletters in the first period. Additionally, interACT progress were reported twice on webinars organized by the CARTRE project. interACT has also established a lively information exchange with its stakeholder group. Journal and conference papers have further raised awareness. Three workshops have been organized on external HMI design and evaluation methodologies. In addition, interACT is engaged in the US-EC twinning with its US twinning project AVintent and is closely collaborating with Japanese partners from the SIP-Adus project. The project is also continuously exchanging information with other projects in this area such as TrustVehicle, BRAVE, L3Pilot, inMotion, @City.
By using the data of the observational studies the project has derived psychological models of interaction applicable to all traffic scenarios targeted by interACT. Additionally, the data allowed us to set up a “catalogue†of both explicit communication means and implicit cues used to understand behavior. Quantitative interaction models will allow the inâ€depth study of interaction scenarios in computer simulations also for other scenarios. Project partners also worked on improved algorithms using data of onâ€board sensors (video, LIDAR, radar and ultrasonic sensors) for intention recognition and behaviour prediction of other road users. The results include a new approach of using V2X technology and mobile phone data as additional input to assess the potential behaviour of other road users and a longâ€term prediction of pedestrian behaviour (> 3sec ) using environment analysis and aâ€priori knowledge from the psychological interaction models. The project work resulted in the definition of three interaction strategies for AV interaction that influences the further design work of the project. Partners further developed s explicit HMI messages in combination with implicit behavioural cues for other road users. With regards to hardware products, the project develops a new product, the 360° lighting system and use case based lighting elements. Our safety solution considers uncertainty and thus does not require any new sensors. A safety layer which always provides a fallâ€back solution for reaching a safe state was developed. This was done by using novel onâ€theâ€fly techniques that drastically reduce the need for testing and thus, the verification costs. Only the safety layer will require certification.
For the upcoming months it is expected that results will be achieved related to the evaluation methodologies. Project partners are planning to deliver experimental setups and test regimes to assess the quality of interaction and cooperation by simulator and real vehicle studies and define quantitative and qualitative metrics to support these assessments regarding acceptability, efficiency and safety. interACT results are significantly supporting the shortâ€term introduction of AVs in mixed traffic environments. All project solutions will foster an increased societal acceptance of AVs via improving easeâ€ofâ€use, road safety and traffic flow. Industrial project partners are enabled to fully exploit project findings, increasing the potential benefit, sale and adoption of AVs. With leading manufacturers such as BMW, BOSCH, CRF and HELLA on board, the project has the ability to ensure results are integrated at a fast pace, allowing Europe to remain at the forefront of this research.
More info: http://interact-roadautomation.eu/.