How might autonomous driving tech be shaped to support mundane and “non-viable” uses?

As with e.g. BoatSense, there are everyday interactions that could be supported using emerging tech. However, the systems of value required means that the access to this technology is unevenly distributed.

All this talk about AI and the current fuss about autonomous vehicles should trickle down to our little community sooner rather than later?

The community is located on a small island not too far from Stockholm city centre. However, the water surrounding the island, and the deliberate lack of tarmac means some logistical challenges arise. Deliveries of goods end up not making it across the water. Small postboxes on the mainland work for smaller parcels that don’t require a signature, but it’s still a hassle to make the trip to see if there is a letter waiting in the postbox or not. Many other factors make for a challenging environment for goods deliveries and this is an attempt to try to negotiate it while exploring self-driving technology.

With limited time and resources, and to not end up knee deep in the basics it was necessary to break things down and use whatever modular tech that doesn’t require an advanced degrees in its respective field. DonkeyCar was an excellent find that helped me get a 1/10 scale RC car driving autonomously reasonably quickly and cheaply. The platform will be a used Permobil C300 as so much is already done in terms of mechanical and control aspects, and it’s not as costly as commercial robotics platforms.

Getting all the parts up and running was a very interesting exercise that allowed me to dip my toes without having to get in and learn ROS at the very deep end. This is still the plan moving forward, as it’s the tried and tested software solution for this area.

Next steps are working with a local roboticists’ club to get a control interface for the Permobil powerchair up and running, mounting sensors and cargo area on the powerchair, build my own GPS-RTK base station for centimeter level positioning accuracy and figuring out how to use a thermal camera for identifying and avoiding people. And digging into ROS.