Spiri Readiness

Hi Folks,

We built and shipped a small batch of Spiris over the last few weeks, but we haven’t got as far into production as we hoped. Technically, we are ready to go. We have been waiting for a tranche of funding from our government and private partners to launch the beta at full scale. In the meantime, we are doing our best to bootstrap.

Along those lines, we are preparing a production test of 20 Spiris, which will go out to Kickstarter backers and research partners. Our aim is to get to a production volume of one Spiri per day built and shipped, and to scale up from there after we bank our coming round of funding.

We have only recently overcome the last hurdle with our camera driver, which has plagued us for months and months. In the image below, you see on screen a live feed from one of Spiri’s cameras (which have been pulled out of Spiri for upgrade work and testing). We have also been working on bug fixes, feature enhancements, and documentation. Here are some highlights:

  • Cameras are now enabled for gstreamer and (for application development) libArgus, which means access to GPU optimizations on the TX1 and TX2;
  • Cameras, by the way, connect directly via CSI rather than USB, which is much better for bandwidth, and frees up your USB port for anything else;
  • Two ways to do autonomous flights: either use “mission” mode on the flight controller, where you upload a pre-written mission (e.g. through QGroundControl), or use “offboard” mode, where you send navigational commands directly from the TX1 or TX2 to the flight controller.

We have created a settings list for the flight controller that works well on Spiri. We also have the ORB_SLAM2 algorithm running, which Spiri can use to map out the features in its surroundings and locate itself within them.

We have run out of technical and manufacturing problems, and we are now sorting out the launch funding. Spiri is ready.

On a flight test about a week ago, I (Patrick) flew Spiri into a telephone pole. It was human error: I thought Spiri was getting too close to the pole for comfort, so I took the controls, and promptly flew it in the opposite direction I intended, straight into the pole. It hit hard, bounced off, and stayed in the air. The top of the shell took some damage, but I was impressed at how well the flight controller reacted, everything from the motors activating their safety stop and restart, to the recovery to stable hover. But don’t tempt fate or do this on purpose!

We have been working towards pilot projects in agriculture, local first responders, construction, and utilities. Right after the telephone pole accident, in fact, we performed a successful demonstration for police, fire and medical responders in the town of Vineland, NJ. Spiri, this time without errant human interference, carried out three beautiful pre-programmed missions. We are seeing more and more interest in using robots for these kinds of applications, and more and more acceptance, at least conceptually, of autonomous flight and easier regulations. It is going to be an exciting time for robotics!


Caroline Glass