User interface (UI) design for modern agricultural machines is crucial to the farmer’s ability to accomplish their operational goals. A UI must address the information needs of the operator and adhere to the principles of human factors and UI design. This thesis describes the development and evaluation of a UI for an agricultural air seeder using a User-Centered Design (UCD) approach. Our primary focus was on the goals and information needs of the operators following situation awareness (SA) principles. We designed, tested and improved UI by conducting laboratory experiments using human participants in a two-phase study.

The first phase of the study involved the design and evaluation of eight elements of the air seeder interface on an individual basis. We compared the new UI elements with existing baseline (also referred as original or old) elements using the metrics of situation awareness (SW), mental workload (MWL), and subjective feedback (SF) of research participants. Comparison of the UCD and baseline UI designs showed improvements in SA, MWL, and SF of the users. The UCD interface elements invoked greater SA (the maximum improvement was 11% with a mean difference of 5.0 (4.8%), 95% CI (6.47, 3.60) and P < 0.0001) and lower MWL (the maximum reduction was 19.7% with a mean difference of -5.2 (-7.9%), 95% CI (0.11, -10.36), and P = 0.0228). Users significantly preferred the UCD elements over the baseline elements during SF. User’s suggestions and the non-subjective outcomes (such as preference about colors and shapes, and performance consequences) led to further design changes.

The second phase of the study involved the design and evaluation of two versions of the UCD interface (UCD1 and UCD2;​ conceptually different in terms of details and complexity) against the baseline UI using the metrics of SA, MWL, SF and user response time under different levels of automation (low and high) and steering type (manual and auto). UI design proved to be a significant main effect on the response time;​ UCD2 interface performed best (120 cs), followed by UCD1 (141 cs) and the baseline interface (174 cs). UI design did not prove to be a significant main effect on SA (level 1, 2, 3, trend or overall).

Regarding the MWL, UI design proved to a significant main;​ UCD2 interface performed significantly better than the UCD1 and baseline UI. Subjective ratings of the UCD interfaces are proved better than the baseline interface. Odds ratio for the UCD2 to baseline was 12.92, 95% CI (9.17, 18.21), and P < 0.001, and for UCD1 to baseline it was 2.84, 95% CI (2.08, 3.87), and P <0.001.

The presence of automation was also a significant main effect on the SA and MWL of the users. SA was better with the high level of automation with a mean difference of 1.34, 95% CI (0.21, 2.48), and P <0.020. A high level of automation also reduced the MWL with a mean difference of 0.61, 95% CI (0.26, 0.95), and P < 0.001. Moderate levels of automation combinations performed better than the extreme automation options (i.e., fully autonomous or manual).