REPORT ON THE TEST OF TWO DRIVERLESS SHUTTLES IN MONTRÉAL PILOT PROJECT ON PUBLIC ROADWAYS FROM JUNE TO AUGUST 2019 SEPTEMBER 2019
Welcome message from author
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
REPORT ON THE TEST OF TWO DRIVERLESS
SHUTTLES IN MONTRÉAL PILOT PROJECT ON PUBLIC ROADWAYS FROM JUNE TO AUGUST 2019 SEPTEMBER 2019
Imprimé le : 20/02/2020 2/17
1 TABLE OF CONTENTS
1 TABLE OF CONTENTS ................................................................................................................................................ 2
2 CONTEXT .................................................................................................................................................................. 3
2.1 PROJECT TIMETABLE ............................................................................................................................................... 3 2.2 PHASES ............................................................................................................................................................... 3
3 PROJECT IMPLEMENTATION..................................................................................................................................... 5
3.1 MAPPING THE ROUTE ............................................................................................................................................. 5 3.2 CONFIGURATION OF OBUS ...................................................................................................................................... 5 3.3 RECRUITMENT....................................................................................................................................................... 5 3.4 NO-LOAD/TRAINING ............................................................................................................................................... 6 3.5 PUBLIC PHASE ....................................................................................................................................................... 7 3.6 COMMUTER INFORMATION ...................................................................................................................................... 7
4 RUNDOWN OF THE EXPERIMENT ............................................................................................................................. 8
4.1 OPERATION OF THE SERVICE ..................................................................................................................................... 8 4.2 AVAILABILITY RATE ................................................................................................................................................. 8 4.3 RIDERSHIP............................................................................................................................................................ 9 4.4 ZENBUS ............................................................................................................................................................. 10 4.5 KILOMETRES TRAVELLED ........................................................................................................................................ 12 4.6 DISENGAGEMENTS ............................................................................................................................................... 13 4.7 ACCIDENTS ......................................................................................................................................................... 15 4.8 MECHANICAL AND SOFTWARE PROBLEMS ................................................................................................................... 15
5 FEEDBACK ON THE EXPERIMENT ............................................................................................................................ 16
6 CONCLUSION .......................................................................................................................................................... 17
VERSION DATE DESCRIPTION
V1 24/08/2019 Creation of the document
Diffusion
ORGANIZATION FOR …
Laboratoire d’Innovation Urbaine Validation
Ville de Montréal Information
Ministère des Transports du Québec Information
Imprimé le : 20/02/2020 3/17
2 CONTEXT
The test of driverless vehicles on the streets of Montréal ended on August 5. As in the experiments done in 2017 and 2018, we gathered considerable data, reactions and findings that
lead us to believe that driverless vehicles could effectively serve to complement active and public transport modes. We are convinced that driverless mobility will influence travel behaviour. The experiments have enabled us to document and share our knowledge to ensure a gradual integration of this technology, while bearing in mind citizens’ needs and expectations related to mobility. Because they are prototypes, these vehicles provide operational data that let us identify elements to improve. We can thus optimize the performance of these vehicles in tandem with our manufacturer.
2.1 PROJECT TIMETABLE
2.2 PHASES
As early as 2017, Transdev organized an experiment in Montréal as part of the UITP
(International Association of Public Transport) Summit. Event participants had an opportunity to ride a driverless shuttle in the heart of Montréal. In 2018, following an agreement initiated by the Régie des installations olympiques and Transdev, a demonstration phase unfolded over a three-month period. Thousands of passengers tried out the shuttles linking the Viau and Pie-IX metro stations to attractions such as the Stadium, Planetarium, the Espace pour la Vie, the Sports Centre, and the Tour de Montréal, and shared mainly positive impressions. The route, planned on a private site, comprised moderate difficulties. The buses navigated amid pedestrians, service vehicles, and cyclists. More than 2,300 people rode the shuttles along a route a nearly 800 m long. In total, the shuttles travelled 1,134 kilometres under the scorching September sun and on snow in December. In fact, from September 10 to December 7, the temperatures varied dramatically, between 30 C and -11 C.
Imprimé le : 20/02/2020 4/17
2.2.1 Phase 2
2019 saw the launch of a more daring project, in collaboration with the Ville de Montréal. The route chosen offered trips between the Olympic Stadium and the Maisonneuve Market. The shuttles thus shared the road with local traffic: cars, delivery trucks, cyclists, and pedestrians. Notably, they also crossed very busy thoroughfares in the district such as Pierre-de-Coubertin, Hochelaga and de Rouen streets. To achieve the project objectives, several steps and installations were required from various authorities.
2.2.2 From a ministerial order to the technology
Because the driverless shuttles used for the project were not yet officially approved, measures were taken with the SAAQ (Société de l’assurance automobile du Québec) to obtain special permission for the experiment, which was granted by the Ministre du transport. To obtain the approval, Transdev had to comply with a few demands by the ministry, after which a ministerial order was issued authorizing the use of public roadways under certain conditions. Following this approval, the remaining steps in the project implementation were set in motion.
The route for the two vehicles was configured and mapped by the expert team at the manufacturer, EasyMile;
Operators were hired, trained and certified. A prerequisite for participation was that the operator hold a class 4 driver’s licence;
No-load test, which allowed various stakeholders to ensure that the shuttles responded
well to the installations following the planning exercise; Ville de Montréal carried out some work to make the route entirely safe and obstacle free,
to allow the shuttles to run properly.
Imprimé le : 20/02/2020 5/17
3 PROJECT IMPLEMENTATION
3.1 MAPPING THE ROUTE
Mapping the route is an essential step to allow the shuttles to navigate independently. This mapping is done using on-board sensors and the location system. The shuttles are equipped with a GPS (Global Positioning System) receiver that lets them find their location. The position in the GPS fix, generally called the “localization” of the shuttle, is first calculated using a signal emitted by the GPS satellites. Once the location is known, the detection of objects surrounding the shuttle using built-in LiDAR sensors can then be superimposed on a
constructed GPS map. This system, called SLAM (Simultaneous Localization And Mapping), can increase the precision of the mapping: the mapping improves the localization and vice versa. After the 3D map of the environment was created, engineers performed the first data entry in manual mode while the shuttle travelled along its route, to record a route that the shuttle should take. This route is smoothed such that the passengers feel as comfortable as possible while riding the shuttle. Driving instructions are thus assigned for each section of the road such as position, speed, acceleration, stops at red lights, and docking at stations. During the first two weeks of the project, an EasyMile engineer carried out this localization and mapping, followed by the route creation, on-site. The results were then presented to Transdev, which validated them before the deployment phase.
3.2 CONFIGURATION OF OBUS
OBUs (On-Board Units) allow the shuttle to connect to the traffic lights, guaranteeing a smooth ride and heightened security during deployments on the open road. These OBUs were connected in collaboration with EasyMile and Electromega, an engineering company with expertise in traffic light connectivity. Connectivity tests between the shuttle and the OBUs were then conducted, upstream of deployment, by the Transdev and EasyMile teams to ensure that the shuttle ran properly.
3.3 RECRUITMENT
As a public transport operator, Transdev employs drivers who were made available especially for
this project. The skills required to operate a driverless shuttle are similar to those required by bus drivers. In contrast, an interest in new technologies is an asset. In addition to greeting passengers and answering their questions, the operator also acts as a technician. This is why training that culminates in a final exam approved by EasyMile is also required. Operators must also hold a class 4 Québec driver’s licence.
Operators were recruited in early May. Several Transdev drivers were thus selected and trained to operate the shuttles safely.
Imprimé le : 20/02/2020 6/17
Operator training
3.4 NO-LOAD/TRAINING
The no-load phase unfolded over a two-week period. The goal was to map the route and produce a reference map that includes all the features of the terrain. The data acquired in manual mode by an EasyMile engineer on-site were then analyzed and
processed by a dedicated team of EasyMile engineers. The behaviour of the shuttle was then defined section by section and tested to ensure that the shuttle followed all the instructions assigned to its controls. For example, acceleration when leaving a stop must be sufficient to “imitate” the way human drivers merge into traffic, but must also be totally safe. The limits of the shuttle were also evaluated during this phase to define a safety/performance threshold on each section of the route. Operators were then trained in the shuttle’s behaviour,
learned about its mechanical limits, safety standards to follow and procedures in case of emergency during training directed by a Transdev engineer. This phase thus required particular attention because it affects future deployment.
Imprimé le : 20/02/2020 7/17
3.5 PUBLIC PHASE
With great pride, the project was announced on June 27 by the main stakeholders: Éric Alan Caldwell, Executive Committee of the Ville de Montréal responsible for Urban Planning and Mobility and the Office de consultation publique de Montréal, Pierre Lessard-Blais, Mayor of the Montréal Borough of Mercier-Hochelaga-Maisonneuve, and Arthur Nicolet, CEO of Transdev Canada. During the project, the shuttles crisscrossed the streets of the district to users’ great satisfaction. Specifically, 3,896 passengers travelled 1437 kilometres in driverless mode, under generally favourable weather conditions.
3.6 COMMUTER INFORMATION
The buses were equipped with a smart phone that let all passengers find their location in real-time using the Zenbus application or the web page. Already used by the transportation networks in France and in Saint-Jean-sur-Richelieu, and by Limocar in Québec, this application specializes in geolocation in the mobility domain.
By letting users locate shuttle stops and find out waiting times before the next arrival of the shuttle in real time, Zenbus reinforces transport accessibility for commuters. A QR code was positioned at each shuttle stop to let passengers consult the web page easily. They could also use the Zenbus application, available on iOS and Android. Note that these apps were not promoted during this pilot project.
Screen capture of Zenbus on
Android
Imprimé le : 20/02/2020 8/17
4 REVIEW OF THE EXPERIMENT
4.1 OPERATION OF THE SERVICE
Service schedules were defined jointly by the Ville de Montréal and Transdev according to the travel needs identified along the route, the schedules of Maisonneuve Market and operational constraints. The route proposed in the first phase of the project linked the different attractions at the Olympic Park and notably the Tour de Montréal, which hosts Desjardins employees. This service was therefore offered at morning and evening rush hour during the week to serve this specific
clientele. Phase 2 of the project had a different mission: it notably aimed to propose a mobility service to local residents along with visitors to the Maisonneuve Market. The organizers thus decided to offer service uninterrupted from 10 a.m. to 6 p.m., seven days a week. After several weeks of operation, new schedules were proposed to the Ville de Montréal in order to reduce the quantity of operators’ work, which was initially quite high. The service schedule, however, remained the same, namely operation from 10 a.m. to 6 p.m., with the first shuttle running from 10 a.m. to 4 p.m., and the second from noon to 6 p.m. The frequency was thus reduced between 10 a.m. and noon and between 4 p.m. and 6 p.m.
4.2 AVAILABILITY RATE
During an experimental project, regular service may be affected by different factors in the field.
As a result, it was not possible to attain the reference service level for the whole duration of the project.
Number of vehicles in service by operating day
0
0,5
1
1,5
2
2,5
201
9-0
6-2
1
201
9-0
6-2
2
201
9-0
6-2
3
201
9-0
6-2
4
201
9-0
6-2
5
201
9-0
6-2
6
201
9-0
6-2
7
201
9-0
6-2
8
201
9-0
6-2
9
201
9-0
6-3
0
201
9-0
7-0
1
201
9-0
7-0
2
201
9-0
7-0
3
201
9-0
7-0
4
201
9-0
7-0
5
201
9-0
7-0
6
201
9-0
7-0
7
201
9-0
7-0
8
201
9-0
7-0
9
201
9-0
7-1
0
201
9-0
7-1
1
201
9-0
7-1
2
201
9-0
7-1
3
201
9-0
7-1
4
201
9-0
7-1
5
201
9-0
7-1
6
201
9-0
7-1
7
201
9-0
7-1
8
201
9-0
7-1
9
201
9-0
7-2
0
201
9-0
7-2
1
201
9-0
7-2
2
201
9-0
7-2
3
201
9-0
7-2
4
201
9-0
7-2
5
201
9-0
7-2
6
201
9-0
7-2
7
201
9-0
7-2
8
201
9-0
7-2
9
201
9-0
7-3
0
201
9-0
7-3
1
201
9-0
8-0
1
201
9-0
8-0
2
201
9-0
8-0
3
201
9-0
8-0
4
Imprimé le : 20/02/2020 9/17
Several operational factors affected service at different times:
Mechanical or software problems, which required vehicles to be temporarily withdrawn from service so that they could be repaired by the supplier’s teams;
Problems related to the roadway, such as badly parked vehicles and traffic restrictions due to work on buildings situated alongside the route;
Constraints linked to operations of a transportation service, such as the absence of an operator for health reasons.
Mechanical and software problems were the main reason for vehicle downtime. The problems that arose are described in greater detail in a subsequent chapter of this report.
For the project overall, the vehicle availability rate was 68%, namely a rate of 87% for the first shuttle and 49% for the second. Although below the rate expected for regular transport service, this rate is satisfactory for an experimental project of this complexity.
4.3 RIDERSHIP
The shuttles were available to the public over a period of six weeks between June 21 and August 4. Nearly 4,000 people boarded these driverless shuttles and experienced this innovative technology.
Ridership per day during the entire project
Different factors affected the ridership volume:
Number of shuttles operating Weather conditions Consumer traffic at the Maisonneuve Market and the Olympic Stadium.
On average, 97 people used the buses on each day of operation. This average fell to 79 people
when only one shuttle was running and 116 people per day when both shuttles were running.
184
70
80
60
101 93
86 82 81
72 81
93
35
98
133
105
71
139 146
52
83 84
108
74
119
75
34
144
109
67
95 92
107
94
132
101
118
158
130
110
0
20
40
60
80
100
120
140
160
180
200
21
/06/
201
9
22
/06/
201
9
23
/06/
201
9
24
/06/
201
9
25
/06/
201
9
26
/06/
201
9
27
/06/
201
9
28
/06/
201
9
29
/06/
201
9
30
/06/
201
9
01
/07/
201
9
02
/07/
201
9
03
/07/
201
9
04
/07/
201
9
05
/07/
201
9
06
/07/
201
9
07
/07/
201
9
08
/07/
201
9
09
/07/
201
9
10
/07/
201
9
11
/07/
201
9
12
/07/
201
9
13
/07/
201
9
14
/07/
201
9
15
/07/
201
9
16
/07/
201
9
17
/07/
201
9
18
/07/
201
9
19
/07/
201
9
20
/07/
201
9
21
/07/
201
9
22
/07/
201
9
23
/07/
201
9
24
/07/
201
9
25
/07/
201
9
26
/07/
201
9
27
/07/
201
9
28
/07/
201
9
29
/07/
201
9
30
/07/
201
9
31
/07/
201
9
01
/08/
201
9
02
/08/
201
9
03
/08/
201
9
04
/08/
201
9
Imprimé le : 20/02/2020 10/17
Ridership per day and per shuttle during the entire project
4.4 ZENBUS
The Zenbus platform was a great success during the project. Zenbus shared its user data at the end of the project. Specifically, Zenbus provided two types of data:
Number of users Number of different people who used Zenbus at least once during the
period Number of sessions Period in which a user was active (new session starting from 30
minutes of inactivity). Zenbus was promoted via QR codes posted at each shuttle stop. In fact, more users visited the platform via the web link than via the application available on iOS and Android. The popularity of Zenbus grew steadily during the project, reaching peak usage in the week of July 22 to 28, with 474 sessions and 247 users during that period.
0
20
40
60
80
100
120
140
21
/06/
201
9
22
/06/
201
9
23
/06/
201
9
24
/06/
201
9
25
/06/
201
9
26
/06/
201
9
27
/06/
201
9
28
/06/
201
9
29
/06/
201
9
30
/06/
201
9
01
/07/
201
9
02
/07/
201
9
03
/07/
201
9
04
/07/
201
9
05
/07/
201
9
06
/07/
201
9
07
/07/
201
9
08
/07/
201
9
09
/07/
201
9
10
/07/
201
9
11
/07/
201
9
12
/07/
201
9
13
/07/
201
9
14
/07/
201
9
15
/07/
201
9
16
/07/
201
9
17
/07/
201
9
18
/07/
201
9
19
/07/
201
9
20
/07/
201
9
21
/07/
201
9
22
/07/
201
9
23
/07/
201
9
24
/07/
201
9
25
/07/
201
9
26
/07/
201
9
27
/07/
201
9
28
/07/
201
9
29
/07/
201
9
30
/07/
201
9
31
/07/
201
9
01
/08/
201
9
02
/08/
201
9
03
/08/
201
9
04
/08/
201
9
G2-31 G2-35
Imprimé le : 20/02/2020 11/17
Number of Zenbus users per week
Number of Zenbus sessions per week
41
69
116
156
186
217
188
30
7
30
30
29
26
30
16
7
0 50 100 150 200 250 300
17 juin 2019 à 23 juin 2019 (Semaine 25)
24 juin 2019 à 30 juin 2019 (Semaine 26)
1 juil. 2019 à 7 juil. 2019 (Semaine 27)
8 juil. 2019 à 14 juil. 2019 (Semaine 28)
15 juil. 2019 à 21 juil. 2019 (Semaine 29)
22 juil. 2019 à 28 juil. 2019 (Semaine 30)
29 juil. 2019 à 4 août 2019 (Semaine 31)
5 août 2019 à 11 août 2019 (Semaine 32)
Utilisateurs Web Utilisateurs App
65
117
194
333
384
410
408
41
14
49
59
81
81
64
56
8
0 50 100 150 200 250 300 350 400 450 500
17 juin 2019 à 23 juin 2019 (Semaine 25)
24 juin 2019 à 30 juin 2019 (Semaine 26)
1 juil. 2019 à 7 juil. 2019 (Semaine 27)
8 juil. 2019 à 14 juil. 2019 (Semaine 28)
15 juil. 2019 à 21 juil. 2019 (Semaine 29)
22 juil. 2019 à 28 juil. 2019 (Semaine 30)
29 juil. 2019 à 4 août 2019 (Semaine 31)
5 août 2019 à 11 août 2019 (Semaine 32)
Sessions Web Sessions App
Imprimé le : 20/02/2020 12/17
4.5 KILOMETRES TRAVELLED
The vehicles travelled 1,437 kilometres in driverless mode during the six-week project. This represents about 36 kilometres per day of operation.
Number of kilometres travelled by a driverless shuttle
Bearing in mind the downtime of one of the shuttles in the first two weeks of the project, the first shuttle travelled 488 kilometres, compared with 949 km for the second.
Number of hours of autonomous driving per day of operation
0
5
10
15
20
25
30
35
40
21
/06/
201
9
22
/06/
201
9
23
/06/
201
9
24
/06/
201
9
25
/06/
201
9
26
/06/
201
9
27
/06/
201
9
28
/06/
201
9
29
/06/
201
9
30
/06/
201
9
01
/07/
201
9
02
/07/
201
9
03
/07/
201
9
04
/07/
201
9
05
/07/
201
9
06
/07/
201
9
07
/07/
201
9
08
/07/
201
9
09
/07/
201
9
10
/07/
201
9
11
/07/
201
9
12
/07/
201
9
13
/07/
201
9
14
/07/
201
9
15
/07/
201
9
16
/07/
201
9
17
/07/
201
9
18
/07/
201
9
19
/07/
201
9
20
/07/
201
9
21
/07/
201
9
22
/07/
201
9
23
/07/
201
9
24
/07/
201
9
25
/07/
201
9
26
/07/
201
9
27
/07/
201
9
28
/07/
201
9
29
/07/
201
9
30
/07/
201
9
31
/07/
201
9
01
/08/
201
9
02
/08/
201
9
03
/08/
201
9
04
/08/
201
9
G2-31 G2-35
0
2
4
6
8
10
12
21
/06/
201
9
22
/06/
201
9
23
/06/
201
9
24
/06/
201
9
25
/06/
201
9
26
/06/
201
9
27
/06/
201
9
28
/06/
201
9
29
/06/
201
9
30
/06/
201
9
01
/07/
201
9
02
/07/
201
9
03
/07/
201
9
04
/07/
201
9
05
/07/
201
9
06
/07/
201
9
07
/07/
201
9
08
/07/
201
9
09
/07/
201
9
10
/07/
201
9
11
/07/
201
9
12
/07/
201
9
13
/07/
201
9
14
/07/
201
9
15
/07/
201
9
16
/07/
201
9
17
/07/
201
9
18
/07/
201
9
19
/07/
201
9
20
/07/
201
9
21
/07/
201
9
22
/07/
201
9
23
/07/
201
9
24
/07/
201
9
25
/07/
201
9
26
/07/
201
9
27
/07/
201
9
28
/07/
201
9
29
/07/
201
9
30
/07/
201
9
31
/07/
201
9
01
/08/
201
9
02
/08/
201
9
03
/08/
201
9
04
/08/
201
9
Imprimé le : 20/02/2020 13/17
4.6 DISENGAGEMENTS
Data on disengagements for the entire project were recorded by EasyMile’s new data monitoring system.
Number of disengagements per day of operation and per vehicle
There are several types of disengagements:
Obstacles – The vehicle’s LiDAR safety sensors suddenly detect an obstacle in the safety zone, triggering an emergency stop (ball, vehicle suddenly merges, etc.)
System – The stop is triggered by the vehicle’s onboard sensors Software – The stop is triggered by the shuttle’s navigation software Emergency stop button– The emergency stop button was pressed by the operator or a
passenger. There may be many reasons for this; they do not result from a decision made by the vehicle.
On average, driverless mode was disengaged every 7.3 kilometres, equal to about every three loops; specifically, every 7.3 kilometres for the G2-031 shuttle and every 6.9 kilometres for the G2-035 shuttle. This average varies depending on the type of disengagement. Disengagements due to obstacles are the most frequent, with one occurring every 20.2 kilometres on average (20.4 for the G2-035 shuttle and 20.0 for the G2-031 shuttle).
Average Distance between two disengagements
0
2
4
6
8
10
12
14
16
18
20
18
-ju
in-1
9
19
-ju
in-1
9
20
-ju
in-1
9
21
-ju
in-1
9
22
-ju
in-1
9
23
-ju
in-1
9
24
-ju
in-1
9
25
-ju
in-1
9
26
-ju
in-1
9
27
-ju
in-1
9
28
-ju
in-1
9
3-j
uil.
-19
4-j
uil.
-19
5-j
uil.
-19
6-j
uil.
-19
7-j
uil.
-19
8-j
uil.
-19
9-j
uil.
-19
10
-ju
il.-1
9
11
-ju
il.-1
9
12
-ju
il.-1
9
13
-ju
il.-1
9
16
-ju
il.-1
9
17
-ju
il.-1
9
18
-ju
il.-1
9
19
-ju
il.-1
9
20
-ju
il.-1
9
22
-ju
il.-1
9
23
-ju
il.-1
9
24
-ju
il.-1
9
25
-ju
il.-1
9
26
-ju
il.-1
9
27
-ju
il.-1
9
28
-ju
il.-1
9
29
-ju
il.-1
9
30
-ju
il.-1
9
31
-ju
il.-1
9
1-a
oû
t-19
2-a
oû
t-19
3-a
oû
t-19
4-a
oû
t-19
VJRD1A10224000031 VJRD1A10224000035
Imprimé le : 20/02/2020 14/17
Software disengagements are slightly more frequent. They occur every 21.6 kilometres (16.5 for the G2-035 shuttle and 50.8 for the G2-031 shuttle). The emergency stop button was pressed every 48.4 kilometres on average, namely every 69.2 kilometres for the G2-035 shuttle and every 30.1 kilometres for the G2-031 shuttle). Lastly, system disengagements are the least frequent. They occur every 49.9 kilometres on average (every 47.2 kilometres for the G2-035 shuttle and every 55.9 kilometres for the G2-031 shuttle).
Distribution of disengagements by type
8,3
55,9
20,0
50,8
31,1
6,9
47,2
20,4
16,5
69,2
0,0 20,0 40,0 60,0 80,0
Ensemble des désengagements
Désengagements "système"
Désengagements "obstacle"
Désengagements "logiciel"
Désengagements "boutton E-Stop"
G2-035 G2-031
15%
36% 34%
15%
E-Stop Button
Obstacle
Software
System
Imprimé le : 20/02/2020 15/17
4.7 ACCIDENTS
No accidents were noted during the project. The obstacle detection system and emergency stop allowed the shuttle to avoid obstacles encountered along the route, be they vehicles, pedestrians or cyclists. Operator prudence, notably during precise manoeuvres of the shuttles (such as at intersections) or in response to careless behaviour by other road users (e.g. motorist suddenly merges into traffic) also provided strong redundancy with the onboard systems.
4.8 MECHANICAL AND SOFTWARE PROBLEMS
Several mechanical and software problems arose during the experiment, specifically during the deployment phase and the operational phase. These problems required intervention by EasyMile’s technical teams in order to put the vehicles back into operation. 1. RIGHT REAR CALIPER BRAKE BLOCKED DURING THE UNLOADING OF THE VEHICLES WHEN THEY
ARRIVED IN CANADA ON MAY 23 (G2-035) The problem arose during the unloading of the vehicles upon their arrival at the Olympic Stadium during the evening of May 23. Transdev and EasyMile tried to solve the problem on-site in order to lower the shuttle from the truck bed. A diagnosis done remotely by an EasyMile engineer identified the source of the problem. The right rear caliper brake was locked, which prevented the wheel from turning. The problem probably occurred during the strapping of the vehicle for transport. EasyMile guided the Transdev team in how to lower the vehicle despite this problem. An intervention on-site was then scheduled for the following days, and the vehicle was restored in preparation for the deployment phase.
2. DOOR MALFUNCTION – MECHANICAL PROBLEM NOTED ON JUNE 22 (G2-031) The problem was caused by a forced opening of the doors. It seems that the doors were opened manually without following the specified procedure, which is to push the red emergency open button. The doors were thus opened with their electric motor. This caused feedback in the electrical circuit of the doors, which blew the fuse of the main power source. In addition, the doors moved out of alignment, causing loss of calibration and divergence between the actual position of the doors and that transmitted to the door encoder. The vehicle was returned to service after the fuse was replaced and the doors were re-calibrated.
Imprimé le : 20/02/2020 16/17
3. SLOWDOWN OF VEHICLES – DIVERGENCE BETWEEN THE SPEED FORESEEN BY THE SOFTWARE AND
THE ACTUAL SPEED NOTED SEVERAL TIMES DURING THE PROJECT – SOFTWARE PROBLEM (G2-035/G2-031) The operators encountered a software problem repeatedly, causing the vehicle to be armed on only one axle. Consequently, the vehicle speed was limited to 1m/s, regardless of the speed programmed for the section. The problem was caused by poor configuration of the traction controller. An intervention on-site was required to reset the controller and return the vehicle to service. This bug was fixed in the last software update.
5 FEEDBACK ON THE EXPERIMENT
This pilot project where driverless shuttles were tested in a dense urban setting, a first in Montréal and in Canada, was a resounding success. All the partners learned a great deal, including Transdev, which was the shuttle operator and the main partner of the Ville de Montréal.
During a pilot project there is no such thing as failure. Problems encountered during the different project phases provide valuable lessons and thus facilitate the future deployment of this type of vehicle. As an operator, Transdev is in charge of the effective daily operation of the project. It is thus a central link between all the project partners, with on the one hand EasyMile, the vehicle supplier, and on the other hand the policy-makers and sponsors such as the Ville de Montréal and the Ministère des Transports du Québec. Ongoing communication between the different partners is thus necessary to deal with unforeseen circumstances, which inevitably arise in a project based on experimentation with a new technology such as driverless vehicles. Good communication, a hallmark of strong reactivity, allows the operator to respond effectively to these contingencies and in turn maximizes the benefits of the project ensuing from all of its dimensions (deployment, operations, technology,
planning, social acceptability, etc.). For Transdev, this project implies adapting our work methods to a technology with many ramifications for daily management of a transport service. It involves mobilizing new skills in all the project phases:
During deployment, by facilitating coordination with the supplier and policy makers.
Strong agility is thus necessary to face contingencies,—and there are many—, along with the exceptional framework of the project, which is not supported by the usual regulatory framework;
During the operational phase, by training employees to use experimental technology that revolutionizes work methods. The workers must master the operation of the vehicle, understand the technology, its capacities and limits, and solve technical problems on the spot. The operator is thus pivotal to the success of an experimental project.
Imprimé le : 20/02/2020 17/17
6 CONCLUSION
This experiment was a dynamic showcase to evaluate the potential of driverless vehicles and their usefulness in the community. Importantly, citizens’ perceptions of eventual access to this
type of vehicle were also gathered. Transdev is convinced that autonomous transport will change our travel behaviour dramatically. Driverless mobility represents a prime opportunity for public transport networks because we believe that driverless public transport services will be deployed before personal self-driving vehicles. Our goal in putting forward the knowledge acquired during these experiments is to foster the gradual and successful integration of driverless technologies in the public transport networks. The Montréal experiment is part of a large series of experiments conducted by Transdev. In total, over 1.6 million kilometres have been travelled, and over 3.5 million passengers have tried the vehicles. The demonstration in Montréal has sparked the interest of transport planners and politicians. In
addition, the owners of sites such as the Olympic Park would like to integrate these vehicles to make their site more accessible.
Related Documents
