Road Tests, Systems, and Training

Designed 15 road test items that simulate driving conditions on Taiwan’s roads to evaluate the environmental response capabilities of self-driving vehicles.

Intersection & Crossroad

Testing the traffic signs and signal detection capability, travel path planning application, and road marking recognition capability of autonomous vehicles.

Movable Street Scenes and Street Furnitures

Testing the shadow of city buildings, street blocks, and the light of street signs on the sensor system (image-sensor) of autonomous vehicles.

Intelligent Bus Shelters

Testing the travel path planning and bus stops of autonomous vehicles.

Roadside Parking lots /spaces

Testing the curb parallel parking, angle parking and reverse parking capability.

Round-about

Testing the roundabout lane recognition, intersection vehicle mergence and travel path planning of autonomous vehicles.

Narrowing driving lanes

Testing the traffic signal/marking detection, and travel path planning capability of autonomous vehicles.

Road curves

Testing the traffic marking detection of autonomous vehicles, lane keeping, following, and travel path planning capability.

Level crossings

Testing the traffic signal detection capability of autonomous vehicles.

T-junction

Testing the traffic sight and signals capability, the application of smart safety system for travel path planning (i.e. V2X), and road marking recognition capability of autonomous vehicles.

Concrete road surface/bridge seams

Testing the environment and traffic detection of autonomous vehicle and the misjudgment of image-based land detection system.

Tunnels

Testing the influence of GPS signal shading, lighting changes and shading conditions on the sensor system.

Tree canopy

Testing the influence of GPS signal and scenarios of lighting changes on the sensor (image) system of autonomous vehicles.

Metal-bridge road surface

Testing the influence of metal bridge on the sensor system of autonomous vehicles.

Extreme Weather

Testing the influence on the sensor system (image) of autonomous vehicles under fog/light rain/heavy rain and other weather.

Back-Lighting

Testing the influence on the sensor system (image) of autonomous vehicles under day/night/front/back light and other lighting changes.

Static Soft Global Vehicle Target

The static soft Global Vehicle Target (GVT) is a 2018 Euro NCAP-approved, 3D soft-foam target for autonomous emergency braking (AEB) testing. The GVT has a soft outer shell that resembles a small passenger vehicle and is built on a soft foam frame. To the sensors of a self-driving vehicle, it appears to be a normal car, as shown in the photo.

The outer shell is fastened to the soft frame. When a self-driving vehicle is moving in a closed space, the GEV can act as a vehicle parked to the front or side. When a collision occurs, the shell of the GEV breaks apart to avoid damage to the test car. The GEV can then be recovered and reassembled for reuse.

Pedestrian & Bicyclist Dummies

The pedestrian dummy and bicyclist dummy with dynamic control systems are good targets for detection by self-driving vehicles. When a collision occurs, the dummy separates from its belt-drive platform to avoid damage to the test car. The dummy can then be recovered and reassembled for reuse.

The pedestrian can carry out animated human leg movements to appear like a person in stride while the bicyclist has rotating wheels to mimic bicycle movements. They can be fully detected by self-driving vehicle systems. Both devices are 2018 Euro NCAP-approved pedestrian targets for Autonomous Emergency Braking – Vulnerable Road Users (AEB VRU) system testing and verification.

Pedestrian Belt-Drive System

The pedestrian dummy and bicyclist dummy are placed on a belt-drive sled that enables them to cross the road in front of a self-driving vehicle. In a collision, the dummy separates from the sled, and the vehicle can run over the sled without causing any damage. The dummy can then be quickly reassembled and reused. The control system includes the following:

1. Controller: Contains a pull motor, a power source supply unit, and a pull module.

2. Sled: Connect the pedestrian dummy or bicyclist dummy to the sled. Movement is carried out by the belt-drive and controller. In a collision, the dummy separates from the sled.

3. Situational Software: Speed is controlled based on situational needs.

Video VBOX Pro

Video VBOX Pro is installed in the test vehicle to provide track location, loop time, speed, acceleration and other information, such as GPS data, UTC time, CAN data or synchronized video. It is equipped with up to four waterproof cameras. During testing, the graphical overlay is produced in real time and embedded within the footage while critical measurements and video are recorded and saved onto a memory card.

VBOX HD2 HDMI

VBOX HD2 HDMI uses two 1080p wide angle camera lenses that shoot 60 fps (or 30 fps for HDMI output). Data are recorded and saved on an SD card. An embedded picture in picture feature shows the film content on an external display during the recording. Like VBOX Pro, when installed in a test vehicle the VBOX HD2 HDMI records GPS data, UTC time, CAN data or synchronized video. The graphical overlay is produced in real time and embedded within the footage while critical measurements and video are recorded and saved onto a memory card.

LiveU Solo HDMI

LiveU Solo HDMI is wireless video encoder that can send VBOX HD2 video to YouTube.

High-precision GPS

The control system uses GPS data to analyze when to trigger motion of the pedestrian dummy and bicyclist dummy. When a self-driving vehicle is moving within the closed facility, GPS is also used to measure data of the vehicle as well as the pedestrian and bicyclist targets (location, vehicle speed, braking deceleration). The system includes the following:

1. Satellite positioning station. The signal range could cover all of the Tainan Shalun closed self-driving vehicle testing facility.

2. Real-time kinematic satellite positioning equipment and gyroscopes provide the location, speed, acceleration, yaw angle speed, and other data of the self-driving vehicle. The data can be recorded in real-time.

3. Wireless communication data broadcast equipment supports differential correction for the (1) satellite positioning station and the (2) real-time kinematic satellite positioning equipment, which enables high-precision GPS measurements.

4. The test vehicle positioning communications system includes a data acquisition processor and a real-time computing unit which articulate the relative positions, distance, and speeds of the self-driving vehicle and the target object. These data are recorded in the data acquisition processor. Wi-Fi-based wireless modules support communications between the self-driving vehicle and the target object.

Electric Control Platform Vehicles

Available for use by academic research organizations conducting self-driving technology R&D or domestic manufacturers of automobile components/electronics. Also used to support initial product development or functional performance verification.

Taiwan CAR Lab provides different levels of service to clients based on their testing or R&D needs. Service planning is as follows:

Level 1: Basic testing or performance comparisons of simple external components. Taiwan CAR Lab fulfills the testing and verification tasks.

Level 2: In-depth testing techniques and products integral to vehicle control, with the opening of some chassis control privileges for electric control platform vehicles. For joint R&D by Taiwan CAR Lab and the testing client, the testing equipment interface is integrated with the control system to enhance overall testing and verification.