Mechanical and Electrical
Resistor Sensors Used & Placement
The resistor sensors added to read the position of the actuators were not completely identical. This caused the start position of the resistors to be 915 and 955 due to their sensitivity.
Lawn Mower itself
Neutral is at neutral before the mower starts moving. Neutral is slow after the mower is already moving. Jerking the handles to the farthest back position while the mower is moving is the break. Moving the handles to the farthest back position after the mower is in Neutral is reverse.
Because of these hugely particular conditions, controlling a zero-turn mower has proven to be quite difficult.
To solve this issue, I bought a linear actuator of the same size, but 4X faster then the current actuators. This allowed for faster reaction time simulating something closer to a human.
Wheels didn’t start moving at the same time
The actuators would be turned on at the same time however, the wheel would not start moving at the same time. This showed the handles themselves on the mower were not perfectly aligned.
Metal Work was not Exact
Throughout the modifications of the mower, I created drawing of the required frame additions. When I brought them to a machine shop, because the shop specialized in welding, the pieces were not perfect. I then modified them to the best of my ability including cutting them, welding them, and adding small spacers to the required places.
Actuators Moved at different speed
The two identical actuators bought did not move at the exact same speed and therefore did not end up at the same length at the same time. This caused the lawn mower to drive crooked. To attempt to fix this issue, encoders and resistor sensors were added; along with correction code. The actuators were also much too slow moving (0.38″/sec). This caused a lot of delay issues. To fix the issue to the best of my ability, the mower itself was set on the slowest speed. This Fixed a large part of the delay but still not perfect.
align the encoder.
metal frame aligning the encoder.
Attempted Fixes:
To attempt to move the mower straight, code was added to read the values of the encoder and would change the actuators to keep the wheels turning at the same speed. This correction code worked for slow speeds, but the linear actuators, could not adjust themselves fast enough to keep the mower going straight. Because of this as the mower drove faster the ‘S’ shaped driving pattern became more exaggerated.
Different code patterns used on the encoder included keeping the encoders at the same number of pulses and ensuring the same amount of pulses happened every 50 ms.
Finally, the left actuators was switched out for an actuator 4 times faster then the original. This allowed the fast actuator to be controlled to match the slow actuators movements creating a straight line.
Improvements:
When building the mower there were many things that were not considered that should have been. One of the main issues, was the wiring of the machine. The wiring was first put together in a huge mess. This used extra time because the machine then required to be rewired afterwards.
Metal Frame Iterations:
First Idea
This first model of potential metal work was going to sit right under the seat allowing for the machine to easily transform from a self driving mower to a ride on mower. This idea was not going to work as it prevented the seat from landing on the springs, the box would get in the way of the driver, and there was a large amount of wasted material in this design.
Second Idea
The second idea was far more material efficient and was used on the conversion of the zero turn mower. This version worked well however, to make the design better, it needs to be a bit longer before the bend to stay out of the way of the parking break and mower level shift stick. Additionally, it should have a cross bar to create a sturdier frame.