Most robots do badly in unpredictable situations. I’ve seen a dexterous robot stop its assembly work because a piece of plastic didn’t land precisely flat. A human worker would have nudged the thing into place without a thought. The robot halted and flashed a red light — bat signal for a human to come save the day.
If robots are to live and work with people outside of assembly lines, they’re going to have to be more flexible. That’s why I spoke with Naganand Murty, co-founder and CEO of Electric Sheep. Electric Sheep (yes, the Philip K. Dick reference is intentional) makes robots for yard maintenance. They mow lawns, edge and trim lawns and bushes, and blow leaves.
That means the company had to develop robots that could handle a workday where no two environments are alike, and where conditions change constantly — from rainy to sunny, empty to kid-occupied, muddy to dry. And then there are those damn hidden tree roots.
Then, too, Electric Sheep has been growing by buying landscaping businesses. So its employees are people who have worked in outdoor maintenance. Unlike, say, autoworkers, they aren’t robot-savvy. They’re civilians, like me. So I was curious how they were getting along with their robot colleagues.
Murty and I discussed those kind of human-robot interaction questions. And the world model and robotics tech he’s worked on to cope with unpredictable landscapes. And why sometimes the worst thing that can happen in a simulation is that it works perfectly. Plus a lot of other interesting issues in 2024 robotics. Give a listen.
A note about the transcript: Substack’s transcription feature is in beta. (That’s why it labels Naganand as “Speaer 3” even though there are only two of us.) I’d be grateful for feedback about how well it works. If it’s not yet up to snuff I can provide transcripts from a different source. If you spot problems, please let me know.
Electric Sheep's Naganand Murty