What We’ve Been Up To

It’s been over a year since we did our last detailed update on what Altius is up to as a company, and while we can’t talk in detail about all of the projects we’re working on, I felt a quick update would be worthwhile for those of you following us on the outside.

DARPA Phoenix Phase 1 R&D Efforts
For most of the past year and a half, Altius has been working on two Phase 1 contracts for the DARPA Phoenix Technologies program. The first of these contracts had Altius leading a team (including Roccor LLC, Ecliptic Enterprises Corp., and a robotics research team at CU Boulder) in the development and fabrication of a prototype deployable camera/illumination system for potential use in the Phoenix mission. The second contract had Altius supporting a JPL robotics team in developing a Gecko Adhesive gripper capable of adhering in vacuum to flat surfaces such as solar arrays or deployed PODS systems. Further details will require me submitting materials to DARPA’s DISTAR system for public release authorization (which typically takes about a month). I hope to put together a more detailed write-up with more pictures and videos at some point for DARPA to review.

In the mean-time, here are a few public release-approved pictures/videos of the STEM Arm Proximity Awareness System prototype that our team delivered to NRL/DARPA back in July:

STEM Arm Proximity Awareness System
5DOF STEM Arm Proximity Awareness System Ground Test Prototype

3DOF STEM Arm Project
In the lull between Phases 1 and 2 of DARPA Phoenix, we decided to initiate an internally-funded hardware project to build-on some of our previous deployable robotics work for DARPA and NASA. Those of you who watched the Youtube video of our Sticky Boom Zero-G flight back in 2011 probably noticed that one of our main challenges was accidentally bumping the target away before the electroadhesive could lock onto it. The thermal-vac testing we did at LM’s Waterton Canyon facility during our NASA SBIR later in 2011 (shown below) demonstrated 10-12lb pull forces on a similar metal sphere, but the contact forces when the gripper first touched the target were enough to accelerate the low-inertia target away from the capture head before Mike Judson could manually trigger the electroadhesives. I think this was a clear case of humans making crappy robots.

We were mulling over what sort of feedback control system to solve this problem when we saw a presentation by NRL at the DARPA Phoenix Industry Day (back in November 2011) talking about the FREND robot arm they developed with MDA-US for the previous DARPA SUMO program. They had been trying to deal with a similar bounce-off problem while trying to pinch-grasp Marman-clamps on a simulated satellite. I can’t find a video online, but they showed the FREND arm trying to grab a 4000kg sled mounted on an “zero-friction” air-bearing table with and without an active compliance control system engaged. When the arm was operating without the compliance control on, the 4-ton target would bounce away before the pinch grasper could close. However, by turning on their compliance control algorithm, they were able to minimize the initial contact impulse, and keep constant contact with the target spacecraft for several seconds–basically the control algorithm allowed them to actively minimize the contact forces, turning the contact into an inelastic collision. The challenge with their existing system was that they were actively controlling the torques on 7 simultaneous rotational joints (at sub 3ms loop speeds) using inverse kinematics, while also running collision avoidance loops on top of that. Long and short was that this system required 7 rad-hard processors to function out at GEO. Here’s an NRL picture of the FREND Arm mounted on a test rig that allows them to test compliance control for spacecraft non-cooperative capture:

MDA-US FREND Robot Arm Being Tested at NRL
MDA-US FREND Robot Arm Being Tested at NRL

A hunch I came up with at the time was that with a STEM Arm, the extendable/retractable degree of freedom may enable you to mathematically control the three force directions and three torque axes independently. The thought that was by decoupling all of those equations, you might be able to drastically simplify the computer controls, enabling a much simpler, cheaper arm for capture applications–whether you’re talking about pinch grasp capture like FREND was designed for, or resettable adhesive capture like we’ve been talking about with Sticky Boom.

We’ll have a newer video to upload in the next day or two, but I wanted to provide a link to a Youtube we posted a few weeks ago of the prototype testbed we put together to start testing these concepts:

Basically we took a lot of the design lessons we learned from the DARPA STEM Arm project, and incorporated them into a motorized system. We used an off-the-shelf thermoplastic-composite STEM from Rolatube for this project. The Rolatube COTS booms aren’t as structurally optimized as the carbon fiber booms Roccor has been working on with us for DARPA, but they’re designed for really rugged use, and they had a version with 8 embedded conductors available off-the-shelf that they were able to ship us quickly. We used a simplified root-support system that’s about halfway between the root support we used for the Rev2 Sticky Boom back in 2011 and the much fancier root support Roccor did for our Phoenix Phase 1 prototype. We incorporated a slip-ring into the STEM boom spooler to enable us to pass power and data between the wrist-mounted sensors/graspers and the base of the STEM Arm. We also developed an Arduino-based controller system that enabled both joystick control and the demonstrate various tip control options. The option shown in the video was using an ultrasonic sensor mounted at the boom tip (with power and analog signals being passed through the embedded wires in the boom) to keep a constant distance away from an approaching object. The newer video shows some of the updated controls/algorithms.

I’ll go into more detail in a later blog post (or two or three), but here are the next steps we envision:

  1. Incorporate position feedback sensors on all the joints
  2. Characterize the system response (including backlash, time constants, etc)
  3. Develop and test a position controller
  4. Integrate some sort of external object tracking sensor (Xbox Kinnect?)
  5. Develop a control loop for tracking a moving object, and dynamically moving the boom tip to match velocities with the target at a small standoff distance
  6. Integrate a force-feedback system, and develop a force-feedback control loop
  7. Demonstrate the ability to track, dynamically approach, and then contact a target on an air bearing table without bounce-off

Obviously there are many intermediate steps, and since this is internally funded, we’re going to be going at a rate driven by available cash/engineering cycles. But we wanted to show you the project and where we’re currently at. If you happen to be in the Denver/Boulder Colorado area, have experience with robotics, and would like to collaborate on this, we’d love to talk with you.

Cubesat Capability Bootsrapping
One of the business areas Altius has been focusing on is how to turn our technology concepts like Sticky Boom into useful services that profitably solve existing, real-world customer problems. While we could just be a developer/manufacturer of custom robot arm and/or Sticky Boom subsystems, that has always seemed like a pretty niche market. One of my epiphanies in this direction came from observing all my friends competing in the cubesat/smallsat-based earth observation market. I know key people at SkyBox Imaging, Planet Labs, Dauria Aerospace, and Planetary Resources. While probably the most important part of their business plans is the fact that they’re all positioning themselves as big data companies that just happen to build their own satellites to gather said big data, what enables that is the belief that cubesats can drastically reduce the cost of developing a “minimum viable product” spacecraft.

This got me thinking about the idea of cubesat-based systems with miniaturized non-cooperative capture systems based on our 3DOF STEM Arm work. At least some preliminary looks at the technologies that are coming available indicates such a system might be possible soon. Here’s an artist’s rendition of our PitBull concept:

PitBull NonCooperative Capture Cubesat
PitBull Cubesat-Based Non-Cooperative Capture Tug Concept

In order to really understand performance and economics for such a system, I wanted to start seeing if I could bootstrap my way into an initial cubesat development capability at Altius. Fortunately with Altius’ proximity to CU Boulder, which has a pretty vibrant cubesat/smallsat program, we were able to find some top notch talent to start working with us on some proposals that could give us the resources to get into this new business area. We are waiting to hear back on our first such proposal, and are working on some additional projects. The hope is that once we have the in-house talent needed to design, build, and fly cubesat missions, we’ll be in a much better position to combine that with our STEM robotics work and pursue commercial, NASA, and DoD cubesat missions that need non-cooperative capture capabilities. Admittedly, this new market area is somewhat of an experiment for us. If we’re unable to make traction with our own internal cubesat development capabilities, we’re friends with several other companies we could potentially collaborate with, as a fall-back plan.

Other Commercial Projects
Other than DARPA Phoenix, one of our biggest business and technical projects we’ve been working on is a commercial-service we’re trying to develop. This one isn’t really related to our Sticky Boom or STEM Arm robotics work, but is more driven by a customer need that a company we’re working with identified for us earlier in the year. Due to some sensitivities with some of the players involved, we’re going to have to hold-off on providing a lot of details, but consider this a teaser that we’ll hopefully be able to spill the beans about down the road.

In addition to that project, we’re also starting to work with some folks regarding terrestrial, mostly DoD applications of some of the STEM robotic arms we’ve been developing. We’ll also hopefully have more to say about those projects down the road. While they’re not directly space-related, they’ll use some of the same hardware pieces and help drive some of the same technologies and manufacturing capabilities.