Morphogenetically Assisted Design Variation (MADV)

Funded by DARPA DSO under the Maximum Mobility and Manipulation (M3) program

Contact Email: madv-team@bbn.com

Troops in theater are often the first to discover an emerging need, and often have creative ideas of how to address it. The ongoing revolution in fast prototyping and lightweight manufacturing technology will soon remove the physical impediments to creating new systems in the field: what is needed are design tools that will allow non-experts, e.g., soldiers, to create variations on existing designs that will better suit operational needs.

Design experts not only know which parts must be changed to accomplish a new objective, but more importantly anticipate the impact this change will have on other components. These are critical skills that non-expert users lack, and they are perfect tasks for a smart design tool. The tool should adapt other components to support and compensate for these changes, and should support rapid fabrication of the final design.

Consider a soldier in the field with a small surveillance robot that can climb over obstacles up to a certain height, e.g., small rocks. The soldier's next mission is in an area that contains larger obstacles that the robot cannot handle, such as tall stairs. In order to accomplish the mission, the soldier needs a way to make a variation of the robot that fits the new requirements. However, he is not a robotics expert.

We aim to fill this gap with Morphogenetically Assisted Design Variation (MADV). A military base equipped with MADV tools running on a computer and connected manufacturing capabilities, such as a 3D printer, would allow the soldier to modify the design and fabricate the necessary parts to create the robot he needs. This design can then be shared with other soldiers and evaluated for addition to general inventory.

In this project, we focus on robotic design, where systems are typically complex and highly integrated, yet relatively small and inexpensive. In particular, we focus on an example robot similar to the iRobot LANdroid, but simpler and less expensive, which we are calling the “miniDroid.” Increasing the speed of such a robot might require changing the size of the motor. This in turn could affect the battery size, the control circuit board, the size of the chassis needed to accommodate all the components, etc. Even small design changes have a ripple effect through many other components of the robot. Finally, since robotics is a particularly challenging domain, progress in this domain may be applicable to other domains.

Software Releases

MADV pre-alpha release, November 16, 2011
MADV pre-alpha release 3, February 12, 2014

Videos

End of year 1 video demonstrating 5x variation for step climbing, wmv format, m4v format
Step-climbing, self-righting, and driving simulations for minidroid (m4v, mov, mp4)
Variant miniDroids: able to climb much larger step (m4v, mov, mp4), high undercarriage clearance (m4v, mov, mp4)
Early stages of body-plan development for a miniDroid robot, including distortion tolerance

Publications & Talks