Time to win
How we helped Friendly Robots cut deployment prep by 60%
Introducing Friendly Robots
Friendly Robots are a small robotics startup in Circuit Launch, Oakland.
They’re all about creating technology that does good things for the world. That innovation is primarily focussed on the janitorial industry, where their fully autonomous, easy-to-use vacuum cleaners are changing the ways we clean.
They have an MVP ready to go – now the hunt is on for a first big contract to prove the business case.
Sunsets and shifting desks
Friendly Robots currently deploy their robot in Silicon Valley Robotics’ office building, home to a cluster of robotics startups.
But because startups in the building move around quite frequently, desk spaces and layouts are ever-changing, making map updating a headache; even high accuracy maps generated offline are useless within a few weeks.
They have an agreement with building managers to run the vacuum robot after work hours, which means it runs in low light conditions.
Small team, big ideas
They’re quite a small team, so they have to focus on tasks selectively.
Tasks such as developing their own VSLAM algorithm are out of the question; it’s just too time-consuming to justify the herculean effort.
So, over to open-source SLAM solutions? Well, not so fast – they’re usually not optimised well enough for computation, they don’t work well in large dynamic environments, or they don’t offer a complete package, leaving the team with no 2D occupancy grids for ground navigation, or ways of modifying maps.
Wrong loop closures meant the SLAM solution Friendly Robots were using sometimes required them to remap. That meant a long demo preparation time. Deployment preparation time would go up as space size went up, making it time-consuming and tedious preparing.
The results
The project has so far been a real success, with Friendly Robots enjoying big time savings.
Thanks to SLAMcore, Friendly Robots were able to decrease their deployment preparation time by 60+% every time they map.
Because the solution was more robust, maps would last between two and three times longer than before.
SLAMcore also prepared a way to modify the occupancy grid which comes in handy with changing environments.
We’re currently working together to integrate wheel encoder to improve localization further.
Check the specs
Want the nitty gritty on our robust, accurate spatial intelligence? Simply fill in the form below to get our fact-sheet in your inbox instantly.
