For the most part the blower does a decent job, although if it sees a taller pile of leaves it thinks it’s an obstacle and circles around it blowing the leaves in every direction. The mower has obstacle avoidance settings that can be changed, but I don’t see it on the blower… am I missing something? It’s making the blower unusable.
2 Likes
I don’t believe there are any really that you can change currently, unfortunately.
@joef985 Hi, Thanks for sharing your experience! The blower relies on both visual sensors and bumper strips to detect obstacles. The visual system is the primary method and cannot be turned off — this is because the blower operates over a relatively long distance, and the airflow is strong enough that it could potentially affect nearby objects, like decorations, small children, or pets.
For taller piles of leaves, the best approach is to run your automatic leaf-blowing plan before the leaves accumulate too high. This way, Yarbo can clear the leaves efficiently without mistakenly treating them as obstacles.
2 Likes
The vision detects shadows in my driveway as obstacles and gives up and stays stuck in the middle of the driveway. Vision needs a lot of improved training. As for understanding nearby objects, it completely misses 24" oak tree trunks in the middle of my yard. And it blows leaves into areas that it’s already cleared (forgetting that those areas are clear) while trying to avoid the giant oak tree. I wonder if the developers understand that leaves come from trees. Trees are in the middle of yards (better avoidance), and trees are on the edges of yards (be able to split the direction). Leaves come down daily (more frequently than snow and grass) which would make scheduling very helpful. To not be able to factor those things in to the usage of the blower, really detracts from it’s value proposition.
4 Likes
Hi @Greg Tall leaf piles and shadows from trees can sometimes be misidentified as obstacles, and we’re sorry for the inconvenience this causes. Our team is continuing to improve the vision model for better detection accuracy.
2 Likes