Learn about our experiences with an autonomous shuttle on public roads in mixed traffic situations in the middle of the City of Zug, Switzerland.
The MyShuttle Project brought an autonomous shuttle into regular traffic in the City of Zug. The chosen routes included speed zones up to 50 km/h and mixed traffic including public transport route buses, regular mixed city traffic, pedestrian crossings, traffic lights and even bike lanes – at the time, one of the most complex environments in Europe for this kind of vehicle.
By deliberately pushing the boundaries, invaluable insights into the operation of autonomous vehicles were gleaned.
Customer acceptance for autonomous shuttles, the integration of flexible offerings into customer information channels and evaluating the technical maturity of the current generation of vehicles were the main project goals.
The research character of this project differentiates it from other autonomous shuttle undertakings in Switzerland. Various hypotheses were tested under real world conditions during the operations in 2019. The limited technical maturity of vehicle reduced the number of project milestones that could be achieved.
Despite various setbacks, the project was a complete success. Of particular note is the progress that was made regarding technical competencies, the homologation process and vehicle operations.
During the complete project timeframe from 2017 – 2019, vast know how was accumulated. On this website, you can find a brief overview of the most essential aspects. A detailed final project report to the Federal Department of Roads (Bundesamt für Strassen – ASTRA) can be found here.
Vehicle make & model: EasyMile EZ10, 2nd Generation
Tare weight: 1100 kg
Length: 3.9 m
Width: 2.0 m
Height: 2.75 m
Maximum approved speed: 20 km/h
1 Safety Operator (standing)
Sensors for autonomous driving.
Localisation LiDARs – front and rear
3D LiDARs – front and rear
Safety LiDARs – front and rear both left and right
The vehicle was also fitted with external facing cameras (front and rear) as well as an internal camera. The cameras were only used for observation in the event of emergency situations.
In order to determine the exact track to be driven, each route first needs to be driven manually, during which the environment will be scanned with the localisation LiDARs. The resulting map forms the basis used for LiDAR-based localization. Together with the GPS, the vehicle can determine its location to within a centimetre. The route is then optimised and a track defined enabling the vehicle to drive precisely along the virtual (railway) tracks.
The accelerometer and the odometer are used as a plausibility check during the journey, ensuring that it doesn’t stray.
One of the goals of the MyShuttle project was to integrate a dynamic, customer booking driven (on-demand) shuttle service, which was seamlessly integrated in the existing public transport system. To achieve this, the following aspects needed to be analysed, tested and continually re-evaluated based:
How can on-demand services be integrated into the existing customer information systems?
How can the multi-directional communication between service providers, infrastructure and customers be achieved?
How can a trip from A to B using multiple transport modes (multi-modal) be offered in a transparent and comprehensible form?
The findings from these questions can best be discussed based on the following process flow.
The result is a process flow, which works well, both from the customer perspective as well as the service provider perspective. It was proven that this process flow could be implemented while leveraging the existing public transport IT systems.
The customer (1) requests transport between two locations via an app (a). In the MyShuttle Project a new MetaRouter component was implemented and tested.
The MetaRouter App (2) shows combined (multi-modal) transport options (b). Multi-modal means that different transport modes can be used and the customer can choose, which combination best meets their wishes.
If the customer selects an option containing MyShuttle, a reservation request is sent to the Fleet Management system (3).
Depending on vehicle and space availability, the request will be transformed into a booking with a trip time and destination which are completely dynamic and based the customer’s transport needs. The trip can then be distribute via the regular public transport customer information systems (4).
The new transport offering (e) can also be displayed to other customers, via the regular digital customer information channels (5).
The following pictures show how the MetaRouter functionality was built into the SBB Preview App for the MyShuttle Project.
The MyShuttle Project explored the boundaries of possibilities with the existing Public Transport Customer Information systems. Where necessary, new approaches were designed and prototyped and the following knowledge generated:
The core customer information systems today are built around unidirectional mass communication. They convey plan and real-time information about static route-based offerings. Dynamic on-demand offerings cannot be fully integrated.
The existing systems rely on pre-defined stop-locations (eg. train stations, bus stops) and cannot currently handle dynamic route definitions. For our testing purposes, we were able to dynamically create new route definitions and distribute this information, however concerns were raise about the scalability of such solutions with the existing system architecture.
Multi-modal trip planning requires the relevant information to be available in a consistent form, despite the inherent differences of the modes of transport. As such, standards need to be established and enforced.
The current public transport laws denote a "transport guarantee", ie. that a particular route will be driven at a given time, but not necessarily that there will be space available for all passengers. Implementing a dynamic on-demand offering as public transport will require refinements to this definition.
Late in 2018 a permit was issued allowing the Shuttle to operate on public roads in mixed traffic. The operations in 2019 were executed on the Routes 1-6, below.
Total distance covered: 2561km
84% of this distance was driven in autonomous mode
The 6 routes were driven in autonomous mode (2a was driven manually until 2b could be driven in autonomous mode)
Maximum speed in autonomous mode: 16km/h
Maximum speed in manual mode: 10 km/h
Total passengers transported: 534 people
To ensure safe and secure operations, the necessary processes were defined and implemented.
MyShuttle was integrated into the standard operating procedures of the SBB Operations Centre.
The Operations Centre was responsible for the remote monitoring of the vehicle and supported the on-board Safety Operator.
The Operations Centre was informed about operational variances (eg. Battery Levels) directly via the software
The Operations Centre was in direct daily contact with the Safety Operators and checked the operational status at the b
The Safety Operators ensured that the vehicle was operated in accordance with the permitted legal parameters.
Relevant events as well as disruptions were documented by the Safety Operators. Appropriate corrective actions were defined by the project team and agree upon by all affected parties.
Various interventions were required in the normal operations:
Automatic Emergency Stops are triggered by the vehicle when it identifies a potentially dangerous situation
Manual Emergency Stops are triggered by the Safety Operator when they identify a potentially dangerous situation and the vehicle hasn’t already executed an Automatic Emergency Stop.
Manual bypass driving occurs when the Safety Operator needs to drive the vehicle, eg. to pass a poorly parked vehicle or around roadworks.
A Soft Stop is a gradual reduction in speed – often until the vehicle completely stops – in accordance with the route being followed, eg. traffic lights, pedestrian crossings. Soft Stops are usually trigged automatically, but can also be triggered by the Safety Operator.
Daily operations usually ran smoothly and with few issues. Heavy rain and snow, as well as airborne dust from building sites caused problems for the vehicle, as did vegetation that periodically grew out into the roadway.
The vehicle was not operated on days with adverse weather, where the performance of MyShuttle would have been adversely affected.
Snowfall as well as falling leaves and pollen affected the reliability of the sensors (making the vehicle react overly causiously)
Static obstacles that were not mapped (eg. building sites and new growth) lead to automatic soft or emergency stops and had to be manually bypassed.
Highly dynamic mavoeuvres (eg. swerving bikes) often lead to Automatic Emergency Stops.
The above graphic shows the categorization of interventions by the Safety Operator.
The new growth of vegetation in spring and summer cause localization issues, as the static mapping was done in the winter months.
Other road users had the biggest impact – particularly cars, bikes and pedestrians in violation of the road rules.
Building works were a great challenge for the mapping and track definition for the Shuttle as they lead to permanent changes in the requirements.
The project had a great focus on various research aspects. Different Use Cases were defined and executed in order to help structure the research and provide a solid basis for the questions to be answered.
Aside from customer acceptance, the main focus of the cases 1-3 was to stress test the complete system. This was achieved by gradually, but continually, increasing the size of the customer group as well as the number of round trips the vehicle drove each day. With case 3, the general public were offered the opportunity to ride in MyShuttle. The cases 4 & 5 were executed in parallel and will be referred to as case 4+ below.
There was a distinctly positive echo from the people who took the opportunity to ride in MyShuttle and the technology was widely accepted. To aid in gathering this feedback, customers were requested to fill in a questionnaire about their experiences after each ride (see the following graphic). The best results were recorded for safety and trust in the Safety Operator. Trust in the vehicle was marginally lower, which can be directly tied back to the technical maturity of the vehicle. The fun factor also rated highly, which is a clear sign of the public interest in autonomous vehicles and their current state of development.
Driving speed rated lowest on average, but with a clear increase in acceptability as the maximum speed was increased over time. Even still, the maximum speed of 16km/h was considered by most to be too low, particularly on the 50km/h streets (Industriestrasse – Route 6).
Future Shuttle Use.
Generally, the customers reported that they were impressed by their experiences and 85% said that they would use such an offering in everyday situations. The feedback was lowest in case 4+ with 78% – most likely because of the App-based booking process that was trialed during this time. The feedback was best during case 3 – the public case – where 92% said they would use such a shuttle in the future.
The breakdown of possible future use cases follows. With these questions it was possible to give multiple answers. There was also a possibility to enter further suggestions where such a shuttle could be used. Some of the answers included: to transport children to and from school, with elderly customers, tourist connections or as an airport shuttle.
Customer & Press.
Below are a few impressions from our test customers, including a video which was produced before the significant driving speed increases in summer.
In addition to tremendous support within SBB (particularly the Operations Centre located at Zurich Airport), the project success could only have been achieved together with our partners and service providers.
"I am extremely proud of what we as a team could achieve and all that we have learned, in such a complex environment."
Project Lead until May 2019, afterwards Marketing & Organisation
"Together with our partners, we have shown that Swiss public transport continues to be at the leading edge of European innovation!"
Operations & Processes
"MyShuttle enabled us to live a dream and promised us a glimse of the future, with less pressure on streets and carparks and an achievable reduction in energy usage. To get there, there are still a number of challenges to be overcome."
Data Architecture & technical Integration
"We make tomorrows solutions possible today."
IT Architect (until 2019)
"Important work to better understand the challenges and potential of in the interaction and integration of the various systems."
Agile Coach and Scrum Master
"Besides the technical challenges, the MyShuttle Project provided the perfect testing ground to push the envelope of agile projects – with all its advantages and disadvantages."
Curious? Would you like to know more about the project or generally about autonomous shuttle services?