Vehicle Technologies for Controlling Risk Factors

ANNEX A:
Terms of Reference

Reference: 2024 REOI 003

 AI-camera-based driver monitoring system
Terms of Reference

Table of Contents

Contents
1 Introduction.................................................................................................................. 3
1.1 Background, Statement of Purpose & Objectives ................................................................... 3
2 Requirements............................................................................................................... 4
2.1 Mandatory Requirements of the Artificial Intelligence camera-based in-vehicle driver
monitoring system ............................................................................................................................... 4
2.2 Optional characteristics of the Artificial-Intelligence camera-based system ........................... 4
2.3 Mandatory ISA’s requirements ................................................................................................ 5
2.4 Optional characteristics of ISA ................................................................................................ 5
2.5 Data Protection Requirements ................................................................................................ 5
2.6 Deliverables ............................................................................................................................. 7
3 UNHCR Responsibilities ............................................................................................. 8
4 Responsibilities of the supplier .................................................................................. 8
5 KPI’s ............................................................................................................................. 8

Page 2 of 8
 AI-camera-based driver monitoring system
Terms of Reference

1 Introduction

1.1 Background, Statement of Purpose & Objectives

Each year nearly 1.3 million people die as a result of a road traffic crashes — more
than 3000 deaths each day—and more than half of these people are vulnerable road
users: pedestrians, cyclists, and motorcyclists. 20 to 50 million more people sustain
non-fatal injuries from crashes. These injuries are an important cause of disabilities
worldwide. 90% of road traffic deaths occur in low and middle-income countries, which
own approximately 54% of the world's vehicles. Unless immediate and effective action
is taken, road traffic deaths are predicted to become the seventh leading cause of
death in the world by 2030, resulting in an estimated 1.9 million deaths each year. The
World Health Organization and the United Nations released a global plan for the
second Road Safety decade of action 2021–2030 with the goal of reducing by 50% the
amount of road traffic fatalities and injuries around the world by increasing activities
conducted at the national, regional, and global levels. One of the requirements of that
global plan is the adaptation of technologies to the road safety system which is also
one of the UNHCR’s goals.

UNHCR has been working on the implementation of countermeasures for reducing the
rate of traffic crashes with injuries and fatalities, and it has been estimated that around
24% of the serious traffic crashes within UNHCR operations are related to
inappropriate or excessive speed. Knowing that there are some technological features
that could help in controlling risk factors, UNHCR is interested in testing add-on artificial
intelligence camera-based vehicle technologies that can help monitoring and
controlling at least the next risk factors related to drivers’ behaviours and Intelligent
Speed Adaptation (ISA) for reducing the number of instances of excess of speed.

Therefore, the main objectives of this initiative are:

1- Test in-vehicle artificial intelligence driver monitoring cameras that allow

identifying risky behaviours.
2- Test Intelligent Speed Adaptation devices (ISA).

The Proof of Concept would be conducted for a six-month period after the technologies
are installed in the vehicles. The devices would be installed in vehicles located in
Colombia in four Toyota Prados. Two vehicles would be equipped with ISA and the
other two would be equipped with artificial intelligence driver monitoring cameras and
the vehicles are located as indicated in the below table.

Technology Location
Artificial Intelligence based camera 1 vehicle in Bogota
systems 1 vehicle in Barranquilla
ISA 1 vehicle in Barranquilla
1 vehicle in Riohacha

The installation of the Artificial Intelligence-based cameras would be conducted in

Bogota and Barranquilla, and the installation of the ISA devices would be conducted
only in Barranquilla.
Page 3 of 8
 AI-camera-based driver monitoring system
Terms of Reference

2 Requirements

2.1 Mandatory Requirements of the Artificial Intelligence camera-

based in-vehicle driver monitoring system
2.1.1. The artificial intelligence camera-based in-vehicle driver monitoring
technology shall comply all the next specifications:
o Drivers’ facial recognition.
o Identifying speeding issues in relation to the posted speed limits.
o Monitoring seatbelt utilization.
o Monitoring distracted driving.
o Identifying driver fatigue-related issues.
o Identifying instances in which drivers are following too close another
vehicle.
o Storing at least 200 hours of recordings.
o Being able to transfer the information stored to a computer system by
blue tooth, LTE and/or WiFi.
o The camera system shall be able to record the driver and outside of the
vehicle simultaneously.
o Functionality for identifying emergency situations. For instance, if a
driver is hit by another vehicle, the driver could press a button to record
the time of the event and facilitate collecting evidence.
o A functionality for establishing speed limits within geofenced areas.
2.1.2. Have a computer-based system for collecting the information of the
variables and being capable of generating automatic reports which should
include a driver’s grading system.

2.2 Optional characteristics of the Artificial-Intelligence camera-

based system
Besides, it’s desirable the system has the below functionalities, and vendors
offering those options would be graded higher.

2.2.1. The camera system with a functionality for blurring the

background, so only the driver should be identified, but it shall not make
recognizable vehicle passengers and other road users like pedestrians,
other drivers, etc.
2.2.2. Compliance with any of the next driver monitoring system
standards: NCAP, GSR, MLIT, GBT, or equivalent.
2.2.3. Anti-hijacking functionality

Page 4 of 8
 AI-camera-based driver monitoring system
Terms of Reference

2.3 Mandatory ISA’s requirements

The ISA shall comply all the next specifications:
2.3.1. Adapt vehicle’s speed according to the posted speed limits.
2.3.2. Have a geofencing functionality allowing UNHCR to stablish its
own speed limits in certain areas.

2.4 Optional characteristics of ISA

Besides, it’s desirable the system has the below functionalities:

2.4.1. Panic button, so drivers could turn it off in case of emergencies.

2.4.2. Driver Identification card reader.

2.5 Data Protection Requirements

Have a computer-based system for collecting the information of the variables
and being capable of generating automatic reports, and complying with
UNHCR’s personal data protection and privacy framework including the below
requirements:
2.5.1. Data Collection and Use:
▪ Transparency and legitimate basis: Clearly inform users about
what data is collected, how it's used, and who it's shared with,
and obtain consent where applicable.
▪ Data Minimization: Collect only the minimum data necessary for
the device's functionality and safety (e.g., speed, location, road
conditions).
▪ Purpose Limitation: Use data only for the specific purposes
outlined and approved by users (e.g., speed adaptation, accident
prevention).
▪ Pseudonymization and Anonymization: When possible,
pseudonymize or anonymize data to reduce identifiability while
maintaining functionality.
▪ Security of Personal: Implement robust security measures for
data security at rest and in-transit, including encryption and
access controls.
▪ Data Retention and Deletion: Define clear data retention policies
and delete data promptly after it's no longer needed.
▪ Users are allocated a personal and unique User ID to ensure
traceability and accountability. Sharing of User IDs is not
permitted.

Page 5 of 8
 AI-camera-based driver monitoring system
Terms of Reference

▪ The segregation of duties model is design effective,

documented, consistently applied, monitored, and periodically
reviewed.

2.5.2. Data Sharing and Transmission:

▪ Secure Communication Channels: Use secure communication
protocols and encryption for data transmission with other devices
or systems.
▪ Third-Party Sharing Restrictions: Minimize third-party data
sharing and only share data with approved partners for specific
purposes with user consent.
▪ Contractual Data Protection Clauses: Establish a data protection
addendum regarding the processing of personally identifiable
information/data.

2.5.3. User Control and Access:

▪ Data Access and Correction: Allow users to access, review, and
correct their data stored in the device.
▪ Privacy Settings and Opt-Out Options: Provide users with control
over data collection and sharing through comprehensive privacy
settings and opt-out options.
▪ Audit Trails and Accountability: Maintain auditable logs of data
access and actions to enhance accountability and security.
▪ All login attempts (successful or not) and logouts are recorded in
an external (to the application) log file, including at least
Timestamp, Origin IP, User ID, results of attempt.
▪ All changes to account privileges and access rights for user
accounts are recorded in a log file.
▪ Hosting provider logs all key traffic and stores it off platform.

2.5.4. Security and Resilience:

▪ Vulnerability Management: Implement regular vulnerability
assessments and patching for the device and its software.
▪ Secure Hardware and Software Design: Prioritize secure
hardware and software development practices to minimize
vulnerabilities.
▪ Incident Response Plan: Have a defined incident response plan
to address potential data breaches or security incidents
effectively.
▪ Regular Penetration Testing: Conduct regular penetration testing
to identify and address potential security weaknesses.
▪ Allow annual data protection audits by UNHCR.

Page 6 of 8
 AI-camera-based driver monitoring system
Terms of Reference

▪ Multi-factor authentication (MFA) (including biometrics as an

option) is required for all end user’s access to the system (where
technically possible) and only disabled based on fully justified
cases.
▪ Application is behind a Web Application Firewall (WAF).
▪ All Personally Identifiable Information is protected in accordance
with the UN Principles (https://unsceb.org/privacy-principles).

2.5.5. Transparency and Compliance:

▪ Privacy Policy and Data Protection Impact Assessment: Maintain
a clear and accessible privacy policy outlining data practices and
conduct regular data protection impact assessments, share with
public versions of Data Protection Impact Assessments (DPIAs)
or results of other due diligence exercise with UNHCR.
Reasonably cooperate with UNHCR to implement data
protection impact assessment requirements where applicable.
▪ Compliance with Relevant Regulations: Ensure compliance with
all applicable data protection regulations, including GDPR,
CCPA, and others applicable to the vendor.
▪ The service to UNHCR is contractually fully logically isolated
from services to all other customers.

2.5.6. Additional Considerations:

▪ Secure storage and deletion of driving history data to prevent
profiling or discriminatory practices.
▪ Implementing explainable AI algorithms for speed adaptation
decisions to promote user trust and understandability.
▪ Engaging with stakeholders, including data protection authorities
and privacy advocates, in device development and data
practices.

2.6 Deliverables
The service provider shall:
• Provide the artificial intelligence cameras and/or ISA devices.
• Install the artificial intelligence cameras and/or ISA devices.
• Facilitate the telematics subscription services for the entire duration of the
pilot project.
• Provide technical support through the entire duration of the project.
• Provide a warranty through the entire duration of the project.
• Provide the computer software for capturing and processing data.

Page 7 of 8
 AI-camera-based driver monitoring system
Terms of Reference

• Provide training in English and Spanish language on the software for

collecting and processing information (the Road Safety Unit can assist with
the translation of the training in Spanish)
• Provide a product manual in English and Spanish.

3 UNHCR Responsibilities

3.1.Coordinating the installation of the devices.

3.2.Analyzing the information provided by the in-vehicle technologies.
3.3.Assessing the performance of the services and the technology.
3.4.Sending support requests when required.
3.5.Organizing training for drivers on the use of technology.
3.6.Organizing a Data Protection Impact Assessment.

4 Responsibilities of the supplier

4.1. Installing the devices. This takes into consideration vehicle replacements too.
4.2. Providing technical support through the duration of the project.
4.3. Replacing defective devices when needed.
4.4. Applying the warranty if necessary.
4.5. Providing training on the installation of the devices and use of the technology.
4.6. Arranging the shipment of the devices to Colombia.
4.7. Do the necessary arrangements for sending the technicians to Colombia to
install the devices.
4.8. Define a data security focal point.
4.9. Report any data protection incidents affecting UNHCR, remediation at their
expense.

5 KPI’s

The service and the technologies will be assessed by measuring the next KPI’s:

5.1. Number of software related problems.

5.2. Number of hardware related problems.
5.3. Number of fake alerts.
5.4. Number of unsatisfactory support services.

Page 8 of 8