Delivering Trust
Autonomous
delivery robots
Case Study: Human Computer Interaction
Autonomous robots are about to transform delivery. These AI-powered machines can navigate streets on their own, promising faster and cleaner deliveries.
The decision
However, ensuring they operate safely and without disrupting pedestrians is crucial. This study will examine the trust in these robots.
The Goal
Determine how trust in delivery is perceived for autonomous robots and the key challenges to their successful integration. The minimum acceptable trust level is 70%.
Project Overview
My Role:
UX Researcher
Team (3):
3 UX Researchers
Duration:
4 Months
Company:
Tallinn University
Year:
2024
Tim Neumann
Martin Perez
Ryan Birmingham
Project Roadmap
Research
Designed a pilot study (6 participants). The survey itself was at least adequate to continue with a full survey with these questions.
Design a full survey study (21 participants). Questions were selected from the possible further questions from the pilot study analysis.
Results
Cronbach’s alpha coefficient
Trust Level Reporting Results
Crosstabs:
HCTS with Background and Demographics
Doubts, Fears and Problems
Features to feel more comfortable
Design Procedure
Methodological Overview
The Objects of Evaluation:
- Delivery robots: https://www.youtube.com/watch?v=U8Iey_wfo0I
Apparatus and Materials: Mobile app or a desktop/laptop
Tools and Methods:
- Study (unmoderated)
- Survey, include 3 groups of questions:
- Background
- Human Computer Trust Scale (HCTS)
- Human Robot Preferences
- These questions target each facet of trust, with different questions targeting the different spheres:
- Perceived risk,
- Benevolence,
- Competency
Participants: 21 Participants, 18-65 yrs.
Active users of various delivery services (e.g., food, medicine, package signing) to holistically represent the delivery space.
Richard
Jorge
Sayuri
+18
Results and Discussions
Trust Level Reporting Results
Cronbach’s alpha coefficient is 0.73
According to Pallant (2013), a Cronbach’s alpha value above 0.7 is necessary to ensure
the reliability of the study’s measurements. In our study, the HCTS scale achieved a value of 0.73, indicating good internal consistency and reliability for measuring trust.
Our findings show low trust levels at 61%, with no trust factors meeting the acceptability scale. With more participants, marginal acceptance in overall trust is possible.
The basic demographic breakdown of participants
Participants aged 20 to 40 showed higher trust in ADRs (average trust > 3.0/5) compared to those aged
50 to 60 (average trust 2.5-2.6/5).
What area do you live in?
People living in urban areas
(average trust 3.19/5) expressed
greater trust in ADRs compared
to those in other regions (average
trust 2.61-2.79/5).
What building/location do you live in?
People living in apartment building (average trust 3.16/5) expressed greater trust in ADRs compared to people living in a single house (average trust 2.74/5).
People who use delivery services
Participants who used delivery services weekly or a few times per week (14/21 participants) had almost marginal trust (3.15/5).
Which is the robot most trustworthy based on its appearance?
Half of the participants (12/21)
preferred robot option 5,
their trust in ARDS (3.06/5)
was lower compared to
those who preferred robots 3
and 4 (average trust
3.29-3.55/5).
Trust in option 1 was the
lowest (2.44/5). Robot option
2 wasnot selected by anyone.
Products received through delivery services
Grocery and restaurant
deliveries are more frequent
(52.4% and 66.7%).
Medicine and fragile items
are less common (28.9% and
38.1%).
90.5% order low to medium
value packages (under 100
USD/EUR, 300 SOL).
42.9% order furniture and
high-value packages.
Concerns and
Suggestions
Robot 5
Doubts and Fears
They majority of users are:
- Afraid people will steal the robot and their item(s).
- Concern battery’s robot turned down in the middle of the delivery.
Most of them ask:
- How do they handle terrain?
- What they do in case the robot
gets lost, breaks or is stolen?
Mitigations Suggestions
- Improve the Robot awareness
of how handle terrain, about
their safety, usage legal aspects. - Adding an access code to avoid theft.
- Happy face with AI.
- 24/7 assistance behind the robot.
- Awareness of clear instructions
to follow in case the robot gets lost, breaks or is stolen. - Guarantees of the service.
Takeaway: SWOT Analysis for a hypothetical company
Strengths
Growing Understanding of robot delivery provides curiosity and may make some people want to try it.
Opportunities
This is an open space, we can find many ways to innovate and provide more related delivery
services.
Weakness
Public trust is not quite acceptable, capabilities and legal situation of the robots is unclear.
Threats
An accident or refusing to reimburse a consumer for a mistake would greatly hinder acceptance at this stage.
Conclusions
- This study revealed mixed opinions on trust in autonomous delivery robots (ARDS). While the trust level was currently at 61%, it was approaching the commonly accepted threshold of 70% for widespread trust.
- Key concerns included delivery safety, user awareness, and confidence in the robots’ performance.
- To build trust, ARDS must improve safety, provide clear user instructions, and ensure reliable performance.
Learnings
- Building trust in autonomous delivery systems (ARDS) takes time. Users start with doubts but gain confidence as they see positive outcomes.
- A smooth, reliable experience is crucial for trust. ARDS must consistently ensure safety and reliability to boost user confidence and adoption.
- Trust in ARDS varies by culture and demographics. Tailoring communication and trust-building efforts to these differences enhances adoption.
Limitations
- This study serves as a valuable starting point, utilizing primary research to explore user perceptions of ARDS.
- To strengthen the generalizability of our findings, future research should involve a larger and more diverse participant pool.
- Employing crosstabulations would facilitate deeper comparisons between different user demographics.
Recommendations
- Expanding the research to include participants from various countries would allow us to consider the influence of sociocultural factors on trust in ARDS.
- Happy face robots have a lot of acceptance between participants. We suggest to keep this feature on them when robots are used for delivery.
- Features and services should empower users by offering control over robot failures, enabling order monitoring, and ensuring security against delivery disruptions.
- Based on these findings, future studies could investigate specific features and services identified as potentially trust-enhancing for ARDS.
