Evaluating the Applicability of GUI-based Steering Laws to VR Car Driving A Case of Width Changing Paths

Evaluating the Applicability of GUI-based Steering Laws to VR Car Driving: A Case of Width-Changing Paths Masahiro Fukui1, Takumi Takaku1, Satoshi Nakamura1, Shota Yamanaka2 1. Meiji University 2. Yahoo Japan Corporation 1

Contribution • Applied two models for width-changing path to VR car driving. • Both models were very accurate. • The model reflecting the gap showed a bit better fit. ≈ 𝑊𝑆 𝑊𝐸 100 m 𝐴 100 m 2

Background • Shortest routes are not always the best. • Calculating driving difficulty from road geometries is necessary. 3

Related work After introducing the narrowing construction, speed was -15 km/h. When getting closer to the construction, speed was -36%. [Sołowczuk+] [Distefano+] Conditions were too few to model the difficulty. [7] Sołow czuk, A.B.; Kacprzak, D. Effect of Lane Narrow ing Implemented on an Urban Dual Carriagew ay on Speed Reduction and Acoustic Environment. Buildings 2022, 12, 31. [8] Distefano, Natalia & Leonardi, Salvatore. (2019). Evaluation of the Benefits of Traffic Calming on Vehicle Speed Reduction. Civil Engineering and Architecture. 7. 200-214. 4

Steering law: models of GUI operation Steering law predicts time to GUI task e.g., menus, scroll bars. [Yamanaka+ ’19] [Accot and Zhai] [Accot and Zhai], [Yamanaka+ ’16] � W � A � � [Accot and Zhai] Johnny Accot and Shumin Zhai, Beyond Fitts’ law : models for trajectory-based HCI tasks, Proc. of CHI ’97, pp.295-302, March 1997. [Yamanaka+ ’19] Shota Yamanaka and Homei Miyashita. 2019. Modeling Pen Steering Performance in a Single Constant-Width Curved Path. In Proceedings of the 2019 ACM International Conference on Interactive Surfaces and Spaces. (Daejeon, Republic of Korea) (ISS ’19). Association for Computing Machinery, New York, NY, USA, 65–76. [Yamanaka+ ’16] Shota Yamanaka and Homei Miyashita. 2016. Modeling the Steering Time Difference Betw een Narrowing and Widening Tunnels. In Proceedings of the 2016 CHI Conference on Human Factors in Computing Systems (San Jose, California, USA) (CHI ’16). ACM, New York, NY, USA, 1846–1856. 5

Previous work: applying GUI model to VR driving Linear in GUI Linear in VR Curve in GUI Curve in VR [Accot and Zhai] [Zhai+] [Yamanaka+ ’19] [Yamanaka+ ’23] � W � A � � • Linear and curved → steering law is applied to car driving. • Width-changing → steering law is NOT applied to car driving. [5] Shumin Zhai, Johnny Accot and Rogier Woltjer, Human action law s in electronic virtual w orlds: an empirical study of path steering performance in VR, Presence: Teleoperators and Virtual Environments, vol. 13, no. 2, pp.113-127, April 2004. [6] Shota Yamanaka, Takumi Takaku, Yukina Funazaki, Noboru Seto, and Satoshi Nakamura. 2023. Evaluating the Applicability of GUI-Based Steering Law s to VRCar Driving: A Case of Curved Constrained Paths. Proceedings of the ACM on Human-Computer Interaction. Vol.7, No.ISS, Article 430, 93-113. 6

Steering law for width-changing paths Two models for width-changing paths have been proposed. Accot+ = 𝐼𝐷𝑁𝑇 = 𝐼𝐷𝑊𝑇 Yamanaka+ > 𝐼𝐷𝑁𝑇 > 𝐼𝐷𝑊𝑇 7

Research question Can these models be applied to VR car driving? Accot+ 𝐼𝐷𝑁𝑇 = 𝐼𝐷𝑊𝑇 Yamanaka+ 𝐼𝐷𝑁𝑇 > 𝐼𝐷𝑊𝑇 8

Experiment settings • The length of width-changing section (𝐴) was 20/50/100 m. • Starting width (𝑊𝑆 ) was 2.5/5.0/7.0 m and ending width (𝑊𝐸 ) was 2.5/5.0/7.0 m. Table of 𝑊𝑆 and 𝑊𝐸 9

Experiment 10

Result: Applicability of models • Both models were very accurate. (N = 20) • The model of Yamanaka+ showed a bit better fit. Accot+ Yamanaka+ 7000 7000 y = 202.1x + 221.66 R² = 0.9695 adjR² = 0.9676 AIC = 258.6757 6000 5000 4000 MT(ms) MT(ms) 5000 3000 4000 3000 2000 2000 1000 1000 0 0 5 10 y = 193.16x + 233.04 R² = 0.9767 adjR² = 0.9736 AIC = 255.8177 6000 15 20 25 30 35 0 0 5 ID 10 15 20 25 30 35 ID 𝑅2 = 0.9695 𝑅2 = 0.9767 *Red plots: MT on widening road Blue plots: MT on narrowing road 11

Discussion: Applicability of models • When ID > 20, the model of Accot+ becomes inaccurate. • Future works on more difficult tasks is necessary. Accot+ Yamanaka+ 20000 20000 18000 16000 14000 MT(ms) MT(ms) 15000 10000 12000 10000 8000 6000 5000 4000 2000 0 0 0 10 20 30 40 50 ID 60 70 80 90 0 20 40 60 80 ID 12

Discussion: Applicability of models • When ID > 20, the model of Accot+ becomes inaccurate. • Future works on more difficult tasks is necessary. Accot+ Yamanaka+ 20000 20000 18000 16000 14000 MT(ms) MT(ms) 15000 10000 12000 10000 8000 6000 5000 4000 2000 0 0 0 10 20 30 40 50 ID 60 70 80 90 0 20 40 60 80 ID 13

Summary Background Modeling driving difficulty on width-changing road is necessary Steering law for drive on straight and curve is proposed Question Can steering law be applied to width-changing roads? Method Two models were tested in VR driving Accot+ Result Yamanaka+ Both were accurate The model of Yamanaka+ showed a bit better fit 14