COTAM — Computable Technology Acceptance Model

A computable model that predicts the Acceptance Rate of future users for the interior of a highly automated (autonomous) vehicle.

COTAM transfers the psychological Technology Acceptance Model (TAM) — trust, perceived usefulness, perceived ease of use, perceived comfort and attitude — into a vector/array form that a design engineer can evaluate numerically. Any interior concept becomes a set of criteria rates between 0 and 1; absent parameters simply contribute 0. The output is a single percentage: the acceptance rate, comparable across concepts.

Mehmet Emin Dilek · Master Thesis, Institute for Automotive Engineering (ika), RWTH Aachen University, January 2023. "Modeling Effects of Interior Design Parameters on User Acceptance Criteria for Highly Automated Driving" (Supervisor: Matthis Hötter M.Sc.; Examiner 1: Univ.-Prof. Dr.-Ing. Lutz Eckstein; Examiner 2: Prof. Dr.-Ing. Peter Urban).

How it works

The user is modelled as a point with 120° of peripheral vision on the seat-rotation axis. The cabin concept is defined in two steps:

1. Scenario-defining inputs — seat rotation 0–360° and seat height 38–46 cm;
2. Posture & equipment — backrest (130–157°), seat pan (11–15°), headrest (95–97°) and the optional components (overhead display, vending machine, fridge, drawers, table, trash bin, lunch server, e-curtain).

Each input is mapped to a user-rated criterion (attitude, perceived usefulness, perceived ease of use, perceived comfort or trust), taken from scientific studies and linearly interpolated between the surveyed points:

The whole model is a cascade of NumPy arrays and standardized beta coefficients (γ′ = β₁ · x, Eq. 4-2):

PU   = [ crit_pu | β_c·pu·crit_c | β_peou·pu·crit_peou ]
ATT  = [ crit_att | β_peou·att·crit_peou | β_pu·att·PU ]
BI   = [ β_att·int·ATT | β_t·int·crit_t ]        Int_act = Σ BI

Criterion (dependent)	Chain
crit_att	seat rotation, seat height
crit_pu	overhead display, backrest, seat pan, headrest
crit_peou	overhead display, backrest, seat pan, headrest
crit_c	facing-seat distance, existence-based components
crit_t	overhead display, existence-based components

Scenario dependencies

• Facing seat — the facing-distance criterion (44–65 cm) is only queried at rotation = 180°, otherwise its rate is 0.
• Overhead display — contributes only when it lies inside the user's peripheral vision; for 160° ≤ rotation ≤ 280° its rates count as 0.
• Existence-based components — included (Y) or excluded (N); excluded means 0.

Beta coefficients (adopted TAM)

Path	β
Trust → Intention	β_t·int = 0.18
Comfort → Usefulness	β_c·pu = 0.343
Usefulness → Attitude	β_pu·att = 0.32
Ease of use → Usefulness	β_peou·pu = 0.02
Ease of use → Attitude	β_peou·att = 0.09
Attitude → Intention	β_att·int = 0.52

To make concepts comparable, COTAM also computes a maximum acceptance value (every array at its best rate) and scales the result (Eq. 4-3, A_min = 0):

Acceptance Rate [%] = (A_actual − A_min) / (A_max − A_min) × 100

Validation

COTAM was validated in the KAI static-driving mock-up of ika (AI-supported assistant for interior development). Experts experienced five scenarios — laptop usage, smartphone usage, smartphone with overhead display, personal audio zone (PAZ) and seat facing with leg distance — and rated them with UEQ, SUS and a trust scale. Feeding the experiment's own criteria rates back into COTAM resulted in statistically significant predictions:

Scenario	Measured vs. COTAM	Difference
Smartphone + overhead display	—	~6 % (7 % without normalization)
Personal Audio Zone (unseen scenario)	—	~2 %
Enriched interior concept	≈ 59.1 % acceptance rate	drop of components possible

Repository layout

.
├── Main Model.ipynb              # the full COTAM computation (entry point)
├── *_PEoU_PU.ipynb / *_Att(Pre).ipynb …   # data notebooks per parameter
├── *.csv                         # measured criteria rates used by the notebooks
├── demo/
│   └── index.html                # standalone interactive simulation (no server needed)
└── docs/
    └── figures/                  # thesis figures, model schematics, flowcharts

Quick start

pip install -r requirements.txt
jupyter notebook

Open Main Model.ipynb — it runs the connected data notebooks and computes the acceptance rate for a chosen cabin concept. The spreadsheets (God of DATA.xlsx, raw questionnaire data) are intentionally not part of this repository.

Interactive demo — just open the file in a browser:

python -m http.server 8000 --directory demo
# http://localhost:8000

References

• Davis, F. D. (1989). Perceived Usefulness, Perceived Ease of Use, and User Acceptance of Information Technology. 319–340. [FRE89]
• Lee, Y., Kozar, K. A., Larsen, K. R. T. (2003). The Technology Acceptance Model: Past, Present, and Future. Communications of the AIS. [LEE03]
• Venkatesh, V., Morris, M. G., Davis, G. B., Davis, F. D. (2003). User Acceptance of Information Technology: Toward a Unified View. MIS Quarterly, 27(3), 425–478. [VEN03]
• Jian, J.-Y., Bisantz, A. M., Drury, C. G. (2000). Foundations for an Empirically Determined Scale of Trust in Automated Systems. IJCE, 53–71. [JIA00]
• Gross, C., Siepermann, M., Lackes, R. (2020). The Acceptance of Smart Home Technology. Springer, 3–18. [GRO20]
• Nastjuk, I. et al. (2020). What Drives the Acceptance of Autonomous Driving? TF&SC, 120319. [NAS20]
• Meyer-Waarden, L., Cloarec, J. (2022). "Baby, you can drive my car": Psychological antecedents that drive consumers' adoption of AI-powered autonomous vehicles. Technovation, 102348. [MEY22]
• Lehmann, S. (2019). How Will We Travel Autonomously? [LEH19]
• Luís Oliveira, L. et al. (2018). Evaluating How Interfaces Influence the User Interaction with Fully Autonomous Vehicle. 320–331. [LUI18]
• Fiorillo, I., Nasti, M., Naddeo, A. (2021). Design for Comfort and Social Interaction in Future Vehicles: Leg Space Between Facing Seats. IJIE, 103131. [FIO21]

Full bibliography in the master thesis (see repository author).

License

Code by Mehmet Emin Dilek. Figures and data stem from the master thesis and the cited publications; please attribute accordingly. No warranty — provided for research and educational purposes.