As a means to further test the effectiveness of the ‘hitTest’ algorithm used with the data glove, several iterations of testing took place. This section will illustrate how testing was carried out. To determine if the haptic simulation was accurate, various types of tests were conducted.
First, users were left to interact with the program as they pleased. Testers were looking at the screen while they interacted with the virtual objects. Once they were done, each tester had to complete a questionnaire where they answered how pleasurable their experience was. So as to avoid any bias in the test, I avoided talking with any of the participants.
The second iteration ran the same exact test; however testers were not allowed to view the screen. When they came in contact with an object, I would ask them if they felt a cube or a sphere. Each time they interacted with an object, I made note of the object they were actually touching and their reaction to it.
The next iteration of the test called for users viewing the screen to interact with virtual objects. The program was slightly changed however. In this case, the function simulated a cube when the user interacted with a sphere and vice-versa. Similar to the first test, to avoid bias, I didn’t communicate with the testers while they were interacting with the data glove.
The next test involved the same program with virtual objects that triggered haptic simulation as well as empty space that triggered haptic simulation as well. Like the blindsided tested, users were not allowed to view the screen. When testers felt they were interacting with an object, they would let me know and their responses were noted down.
As an attempt to avoid bias on my part, an advisor suggested that I run a test at random from the previous two iterations. The last iteration of testing called for running random tests where neither the tester nor I knew which version of the program was running until the test was over. User reactions were noted down and at the end of the test, which version of the program was also taken into account.
First, users were left to interact with the program as they pleased. Testers were looking at the screen while they interacted with the virtual objects. Once they were done, each tester had to complete a questionnaire where they answered how pleasurable their experience was. So as to avoid any bias in the test, I avoided talking with any of the participants.
The second iteration ran the same exact test; however testers were not allowed to view the screen. When they came in contact with an object, I would ask them if they felt a cube or a sphere. Each time they interacted with an object, I made note of the object they were actually touching and their reaction to it.
The next iteration of the test called for users viewing the screen to interact with virtual objects. The program was slightly changed however. In this case, the function simulated a cube when the user interacted with a sphere and vice-versa. Similar to the first test, to avoid bias, I didn’t communicate with the testers while they were interacting with the data glove.
The next test involved the same program with virtual objects that triggered haptic simulation as well as empty space that triggered haptic simulation as well. Like the blindsided tested, users were not allowed to view the screen. When testers felt they were interacting with an object, they would let me know and their responses were noted down.
As an attempt to avoid bias on my part, an advisor suggested that I run a test at random from the previous two iterations. The last iteration of testing called for running random tests where neither the tester nor I knew which version of the program was running until the test was over. User reactions were noted down and at the end of the test, which version of the program was also taken into account.
No comments:
Post a Comment