The Moses Biological Computation Lab at the University of New Mexico has launched the second annual NASA Swarmathon. The Swarmathon is a competition designed to engage students in developing cooperative robotic algorithms to revolutionize space exploration. The students are developing and testing algorithms that could one day be used in NASA’s Mission to Mars. Robots are needed for “In situ resource utilization” (ISRU) by collecting materials such as water ice and useful minerals from the Martian surface to be used by astronauts who land on Mars in future missions.
In its inaugural year, 424 undergraduates from 24 Minority Serving Institutions (MSIs) participated in the Swarmathon. Over 600 students from 40 schools are expected to participate in 2017. The Swarmathon includes virtual and physical competition tracks. Teams participating in the physical competition are shipped kits to build their own “Swarmie” robots. Students write code in to run autonomously onboard the robots in Robot Operating System (ROS) so that robots search, pick up objects and return them to a central collection zone. Students upload their code to identical robots at Kennedy Space Center for the Physical competition in April 2017. Virtual teams develop novel distributed search and communication algorithms for Swarmies running in the Gazebo simulation environment. Virtual and Physical teams compete for cash prizes of up to $5000 for robot teams that autonomously collect the most objects.
The Swarmathon is a cooperative agreement between the NASA Minority University Research and Education Program (MUREP) and the University of New Mexico in partnership with the NASA KSC Swamp Works.
Meet the Swarmies
Swarmies are small robotic vehicles measuring approximately 30 cm x 20 cm x 20 cm. Each Swarmie is equipped with sensors, a webcam, GPS system, and Wi-Fi antenna. They operate autonomously and can be programmed to communicate and interact as a collective swarm.
Inspired by Ants
The Swarmies were designed through a collaboration between the Moses Biological Computation Lab at UNM and NASA Kennedy Space Center’s Swamp Works Facility. At UNM, Dr. Melanie Moses and her students studied the behaviors of ants foraging for food. They then translated behaviors into adaptive search algorithms which are then loaded onto iAnt robots to search for resources.
iAnts are the iPod controlled little sisters of the Swarmies
iAnts emulate ways that ants move, return to remembered food locations and communicate. Foraging algorithms are tuned by an evolutionary process that tailors those behaviors into error-tolerant, flexible and scalable robot foraging strategies in varied and complex conditions. UNM partnered with NASA to apply these search algorithms to space exploration missions which will require the identification and retrieval of resources on the moon, mars, or other extraplanetary surfaces.
Swarmies potential for Space Exploration
Successful exploration of the Moon, Mars, and asteroids requires the location and retrieval of local resources on extraplanetary surfaces. Technologies are needed to find and collect materials such as ice (convertible into liquid water, hydrogen fuel and oxygen to support human life) and rocks, minerals and construction materials to build human shelters. This is referred to as In-situ resource utilization (ISRU). Swarmies present the potential to dramatically improve the ability for robots to efficiently locate, identify and collect resources over large and previously explored territory and further ISRU efforts.
Objectives of the Swarmathon
The goal of the NASA Swarmathon competition is to develop integrated robotic platforms that improve resource retrieval rates by 2–4 fold, compared to the same number of robots operating without cooperation, and orders of magnitude faster than solitary robots. For example, 20 Swarmies can travel and search 42 km of linear distance in 8 hours without recharging; that’s the distance covered in a marathon, and the same distance traveled by the Mars Exploration Rover Opportunity in 11 years.
In addition to being the most effective way to scour large territories for resources, robotic swarms are more robust, flexible, scalable than monolithic robots operating alone. This nascent swarm robotics technology can be vastly improved by combining new algorithms, novel hardware and sensors, and traditional computational techniques for search, learning and data aggregation.