@@ -41,16 +41,19 @@ Going backward on its back is expected behavior, as the sensor in this situation
...
@@ -41,16 +41,19 @@ Going backward on its back is expected behavior, as the sensor in this situation
Going backwards on its front is however not expected behavior, as the robot should in fact try to catch its weight by driving forward as it does until the sensor hits the ground. When leaning on the floor the sensor must therefore see a high raw value which means that it doesn't see any reflected light and thinks that it is in fact leaning backwards instead of forward and thereby trying to go backwards.
Going backwards on its front is however not expected behavior, as the robot should in fact try to catch its weight by driving forward as it does until the sensor hits the ground. When leaning on the floor the sensor must therefore see a high raw value which means that it doesn't see any reflected light and thinks that it is in fact leaning backwards instead of forward and thereby trying to go backwards.
#### Testing environmental influence
With the goal of obtaining a better understanding of the significance of the physical environment, we then tried to take Hurbain's reccomendations for environmental conditions [1] into account. According to Hurbain, the NXTway requires the following conditions to be satisfied:
With the goal of obtaining a better understanding of the significance of the physical environment, we then tried to take Hurbain's reccomendations for environmental conditions [1] into account. According to Hurbain, the NXTway requires the following conditions to be satisfied:
1. The lighting must not interfere with light sensor. Hurbain suggests trying out the robot in a dark room or a room lit only by fluorescent light.
1. The lighting must not interfere with light sensor. Hurbain suggests trying out the robot in a dark room or a room lit only by fluorescent light.
2. It works best on a clear but non uniform surface. According to him, certain patterns give the robot something to "zero" on.
2. It works best on a clear but non uniform surface. According to him, certain patterns give the robot something to "zero" on.
3. The robot must be perfectly balanced when the run button is pressed, as the light level measured at that time determines the equilibrium position.
3. The robot must be perfectly balanced when the run button is pressed, as the light level measured at that time determines the equilibrium position.
#### Testing condition 1: A dark room without windows
##### Testing condition 1: A dark room without windows
We tested the robot in the bathroom where the light could be totally switced off.
We tested the robot in the bathroom where the light could be switced off.
When we first placed the robot in the middle of the room and calibrated it, support it using our hands, it fell over almost immediately.
At first we placed the robot in the middle of the room, started the non-modified program, calibrated it with support from our hands. This attemp made the robot fell over - and thereby not balancing - almost immediately as seen in video 2.
![Robot in dark room]()
*Video 2: The robot in a dark room to test Hurbains first condition for obtimal light sensor use.
##### Including condition 3: Using support during calibration
##### Including condition 3: Using support during calibration
After the first, failed attempt, we tried calibrating the robot while it was resting against the door to the bathroom, as the door was perpendicular to the floor and flat. This allowed the robot to calibrate at a more steady, upright position - this, for instance, reduced the risk of us accidentally tilting the robot when pushing the start button after calibration (or actually during, since it goes wrong because the robot is still calibrating while we are trying to push the button). We placed the robot at a small angle to the door (see Video 2), as we assessed that the robot had too much weight in the front because of its sensor. This worked better, but the robot seemed to be clinging to the door.
After the first, failed attempt, we tried calibrating the robot while it was resting against the door to the bathroom, as the door was perpendicular to the floor and flat. This allowed the robot to calibrate at a more steady, upright position - this, for instance, reduced the risk of us accidentally tilting the robot when pushing the start button after calibration (or actually during, since it goes wrong because the robot is still calibrating while we are trying to push the button). We placed the robot at a small angle to the door (see Video 2), as we assessed that the robot had too much weight in the front because of its sensor. This worked better, but the robot seemed to be clinging to the door.
