Servo Magazine ( December 2015 )
BioGears — It’s Alive!
By Bryan Bergeron View In Digital Edition
Every Sci-Fi enthusiast knows that real robots of the future look, act, and experience the world like their human creators. Witness David in Prometheus and Ava in Ex Machina. Getting to this stage of evolution involves a lot more than internal sensors of joint position. It requires a humanoid physiology — realistic responses to running up a flight of stairs, for example — as reflected in breathing rate and depth, heart rate, systolic and diastolic blood pressure, body temperature, and energy expenditure among many other physiologic parameters.
Well, getting your hands around humanoid physiology just got a lot easier with the release of BioGears — a free open source physiology engine developed with government funding (www.biogearsengine.com)1. This open source extensible human physiology engine is available under the Apache 2.0 license and is initially aimed at game developers, human simulator developers, and educators. BioGears can be used as a stand-alone engine or integrated with simulators and sensors. I am fortunate to be part of the development project, now in its second year.
What’s even more exciting about BioGears from a robotics perspective is that engineers at the primary government contractor — Applied Research Associates, Inc. (ARA, www.ara.com) — ported the entire engine to a Raspberry Pi. The install files are available for free download from the BioGears’ site. Simply follow the install instructions, hook up a few sensors, and you’re good to go. For example, give your robot a virtual heart attack, and require that the nearest human start chest compressions. Too soft, and the robot dies. Too hard, and you have a dead robot and some virtual cracked ribs on your hands.
The team envisions BioGears as part of an affordable, modular humanoid simulator that can be used to teach medics how to handle hemorrhages, blast injuries, and the like, all without placing real humans or animals at risk. Such humanoid simulators exist today, but they’re proprietary and expensive. The BioGears team aims to change all that.
To illustrate just how easy and affordable it is to get BioGears up and running, ARA colleagues, Rodney Metoyer, Zack Swarm, and David Byrd developed a CPR simulator with a $10 analog bathroom scale, a tiny bar magnet, a $7 magneto-resistive sensor (NXP Semiconductor KMA210:115 from Digi-Key), a Raspberry Pi running BioGears, and a few inches of tape. With the BioGears’ simulator programmed to present a humanoid with cardiac arrest, the medic has to assume the correct position over the scale and provide chest compression. That is, the medic has to push down on the scale with just enough force and with the correct rhythm or the simulator will flat line.
What I appreciate about their demonstration is the way they converted an inexpensive analog readout to an analog voltage that can be fed to the A/D front end of a microcontroller. They glued the tiny bar magnet to the dial on the scale and mounted the miniature magneto-resistive sensor on the dial glass, directly over the dial. The same setup could provide non-contact indication of wheel rotation on your next robotic project.
What you do with BioGears is really limited to your imagination. You can extend or modify the existing models or create your own with whatever constraints you like. Perhaps you want a robot that mimics the dual-heart physiology of Dr. Who. You can probably imagine how such an engine could be applied to a combat robot with laser tag sensors. Shoot the robot in the chest, and it stops (bleeds to death) after a minute. Hit it in the leg, and it might limp around for 10 minutes before it succumbs to blood loss.
It will likely be some time before we have (or want) a David from Prometheus, but BioGears is clearly one step in the evolution to realistic humanoid robots. SV
1 BioGears® is funded by the Defense Medical Research Development Program (DMRDP) and administered by the US Army’s Telemedicine & Advanced Technology Research Center (TATRC), Medical Modeling and Simulation Innovation Center (MMSIC) under USAMRMC award number W81XWH-13-2-0068.