IEEE Robotics & Automation Magazine - September 2018 - 94

multiple autonomy requirements into a
coherent robotic system and the challenges
of handling uncertainties imposed by a
natural environment. In this article, we
describe the design of an autonomous foraging robot named Cataglyphis (Figure 1)
that won the NASA Sample Return Robot
(SRR) Challenge in 2014, 2015, and 2016.
This article's main goal is to share the
thinking process behind some of the key
design choices made during our participation in the SRR challenge.

To ensure that technologies developed
during the SRR challenge were relevant to
future space applications, several constraints were put in the competition rules.
Devices that relied on Earth-based infrastructure (e.g., global positioning systems)
or Earth-specific environmental conditions
(e.g., magnetometers and open pneumatic
devices), were not allowed. During the
level 2 challenge, the robots were required
to start outside of the competition field and
Figure 1. The Cataglyphis Rover
move atop a starting platform to one of the
approaching a precached sample
three randomly chosen locations on the
during the 2016 NASA SRR
Background
field, each with different initial heading
Challenge. (Photo courtesy of NASA.)
The SRR challenge was initiated by NASA's
angles. Attached to the starting platform,
Centennial Challenges program to develop
each team was permitted to supply a homrobotics technology to support a future Mars sample return ing beacon that could assist with robot navigation.
mission [1]. A US$1.5 million prize purse was allocated for the
The samples on the field were classified into three categories:
SRR Challenge, which included two competition levels. The 1) four easy samples worth one point each, including the prelevel 1 challenge required the robots to autonomously collect a cached sample and three painted rocks, with properties dissample from a known location on the field and return it to the closed to the teams within the competition rules; 2) three
starting location within 30 min. This precached sample was intermediate samples worth two points each that were placed on
used to simulate stored samples of scientific significance from display for the teams 24 h prior to the challenge; and 3) three
a previous rover. The level 2 challenge required the robots to hard samples worth five points each, with only partial informareturn in two hours up to nine additional samples from tion (e.g., a library of potential inscribed patterns) released to the
unknown locations in a large field, along with the precached teams but no display of the actual samples before the challenge.
sample. During the five-year (2012-2016) history of the SRR
In addition, sterile handling of the samples was required,
Challenge, more than 50 teams participated, and seven teams meaning samples that came into contact with each other durcompleted the level 1 challenge. Cataglyphis passed level 1 ing a competition run would not be counted toward the score.
during its first year of participation (2014) and became the Other than the precached sample, whose location was known,
only robot to successfully complete the level 2 challenge the remaining nine samples were randomly scattered on the
in both 2015 and 2016 for a prize of US$100,000 and field. In 2016, regions of interest (ROIs) were provided to help
US$750,000, respectively.
constrain the search areas, as shown in Figure 2.
Precached
Sample

Intermediate
Sample

Hard
Sample

Figure 2. A satellite image of the competition field, ROIs (purple circles), Cataglyphis' perceived paths (black line from the primary
navigation filter, red line from a simultaneous localization and mapping algorithm), and samples collected during the first 80 min of
the 2016 SRR Challenge. The small blue square indicates the location and orientation of the starting platform. (Background image
©2016 Google Earth.)

IEEE ROBOTICS & AUTOMATION MAGAZINE

september 2018

IEEE Robotics & Automation Magazine - September 2018

Table of Contents for the Digital Edition of IEEE Robotics & Automation Magazine - September 2018