When I figured out the right solution I wish I could say I thought "of course that's correct" but instead I was always thinking "that worked!?" But honestly I loved the jank. It was fun to watch the robot take an unexpected path, which happened often at first. My process of figuring out the solution was fuzzy but not entirely aimless, so it was at least a little more rewarding than blind guessing. For the most part I didn't have to read the instructions but the one thing I couldn't intuit was the condition for when the game would decide to move on to the next input.
I have a suggestion for this scenario:
Suppose the input sequence is <Up>, <Right>. I think it's reasonable to expect the robot will jump up and to the right. Instead it just jumps in place over and over. I think this is because the <Right> instruction is not accounting for the wall it's already touching. If the player is already touching a wall, maybe the <Right> instruction should only terminate after the next wall it touches.
Good job! Love this idea and hope you develop it further.