DARPA ASSIST

Providing soldiers with an in-field, mobile solution for capture, access and automatic sharing of audio, video and contextual data such as soldier state and location. Soldiers wear a Nokia N95 on the chest which streams video wirelessly to an arm worn Zypad WL1100. Additionally a HOLUX GPSlim 236 provides realtime GPS. The software has been developed in C++, C and Python using Pygame/SDL for GUI elements.

Features of the system

• Continuous video capture
• Surreptitious still image capture
• Voice annotation of images and locations of interest
• Mapping of user and fellow soldiers in realtime
• Annotation of images and map (supports undo)
• Immediate sharing of still images & annotations

Responsibility

I built and maintained the in-field system that allowed for real-time media capture, annotation and sharing. Our system was built with off-the-shelf equipment allowing for flexibility of deployment. The system went through three rounds of iterative development and evaluation. Each evaluation round included mock military exercises by GT faculty and staff on campus as well as evaluation with actual soldiers at a military testing ground.

In our final evaluation we scored 80% or higher on six of the seven criteria used by NIST when surveying the soldiers involved in the evaluation. We scored a 100% on the criteria, “If the system were hardened for field use and worked flawlessly, would you use it in the field?” In addition to the NIST evaluations, we scheduled an extra day of evaluation during which the soldiers performed three vignettes. Each vignette was followed by a focus group where we discussed future improvements to the system along with any issues they had with the realistic nature of our vignettes. While we were not able to re-enact emergency scenarios, such as explosions or ambushes, we did cover other missions they commonly undertake, such as humanitarian and intelligence gathering missions.