Mobility Testbed Rover Completes First 100-Meter Drive Test

Astrobotic’s rover development group at Carnegie Mellon University is developing the Mobility Testbed – a rover to facilitate testing representative of the rover that will fly on Astrobotic’s mission to the Moon’s Lacus Mortis region. The Mobility Testbed has the same mobility configuration as the “protoflight” rover that will undergo environmental testing during the Milestone Prize Accomplishment Round but is constructed primarily out of terrestrial-grade components and materials. This reduces cost and lead times while maintaining the physical attributes needed to conduct mobility tests representative of the eventual flight article. Most recently, the team has integrated hardware modifications such as improved motor controllers, main computers, functional test radios, and an all-new avionics bay. Recent software modifications include implementing an SPI motor library, UART communications, a specialized motor state machine, and a custom operator interface. These modifications make the Mobility Testbed analogous to both the protoflight and flight rovers in hardware as well as software.


After integrating these modifications, the team conducted a 100-meter drive test to verify and exercise functionality. The rover was teleoperated via radio from a fixed command station and drove 120 yards along the sideline of an American football field with an average speed of 2.5 cm/s, exceeding its 100-meter goal by 10%. This traverse was completed without incident as expected. The rover never halted and needed only minor manual course corrections. See video of the drive test here.

Map Registration Sensor Package

Map registration is a technique that matches (“registers”) a location in an image to the identical location on a map. Astrobotic’s landing technology registers high-quality, real-time camera images taken during the lander’s descent with terrain maps derived from orbital imagery to achieve safe landing at a precise destination. This capability is key for Astrobotic’s 2015 mission to the Lacus Mortis pit: it enables Astrobotic’s Griffin lander to fly directly over the pit, capturing close-up views of the floor and walls that cannot be seen from orbit or the ground, and then to land within 100m of the pit.

Traditionally, lunar landers have descended blind, using radio to determine their approximate trajectory during descent. They followed a pre-planned trajectory to a final descent altitude near the ground. At the final descent altitude, the computer or a human pilot used RADAR to refine the altitude and velocity estimates and gently touch down on whatever happens to lie beneath them. Because of uncertainty in radio localization and variation in lunar gravity, landing using radio and RADAR alone is accurate to approximately 10km. While this accuracy is good enough for landing in flat, safe regions of the Moon, much greater precision is necessary to safely land near a lunar pit or on the rim of a crater.

Map registration acts in place of radio orbit determination to provide better than 100m landing accuracy. As the lander descends, on-board cameras image the surface of the Moon. These images are registered to maps from the Lunar Reconnaissance Orbiter so that position and attitude can be triangulated. Velocity is then measured as the rate of change in the triangulated location.

Over the next four months, Astrobotic will conduct a series of tests to evaluate the performance of its prototype landing sensors and software. One of these tests is to evaluate the map registration accuracy of the proposed sensor package.

The following graphic shows the standalone sensor package that will be used for this testing. The package contains a camera and an IMU (inertial measurement unit) for gathering the data necessary for map registration. It also contains a GPS and compass – which only function on Earth – for comparison to measure the accuracy of the map registration algorithms.

Camera Box 1
Camera Box 2
Camera Box 3

The dome-shaped element on top of the sensor box is the GPS. The cut away shows the power board, IMU, compass, and camera. The camera is a machine vision board-level sensor with an attached lens. The camera is relatively low resolution (1.5 MP) with fast frame rate and low distortion, and is optimal for capturing images during descent from 20km to 100m. This sensor package could be used for high-resolution mapping on Earth or the Moon. The software can operate entirely without GPS, which would be a boon for terrestrial-based UAVs.


Astrobotic will collect data with this sensor in a series of helicopter and fixed wing flights to evaluate the accuracy of the system at a range of altitudes. A culminating test will integrate the map registration sensor with LIDAR and visual odometry cameras to evaluate accuracy for all phases of landing.

Resource-Aware Planning for Robot Operations at the Lunar Pole

Satellite observations have shown the likely existence of water ice at the lunar poles. Confirmation and exploitation of lunar ice will transform space exploration by providing fuel to support far-reaching exploration and enabling commercial endeavors. Because ice and other volatiles are trapped in lunar regolith, confirmation will require driving and drilling in the hostile environment of the lunar pole. Extreme temperature swings between sunlight and shadow mean that a rover may not survive more than one lunar day, or 14 Earth days, which is a very short time compared to the multi-year treks of rovers on Mars. With such a short operating window, every minute matters. Any decision to stop and survey for ice may mean that there will not be time to survey somewhere else later. A polar prospecting rover also faces other limitations, such as battery power. Solar panels can recharge batteries, but unless a robot’s movements are carefully planned, deep, cold shadows that shift with time could trap it without power and end its mission.

Under a Small Business Innovation Research (SBIR) award, Astrobotic and Carnegie Mellon University are building a computer-aided mission planning tool to improve understanding of the trade-offs between mission goals and resource constraints. This mission planning software may be demonstrated as part of NASA’s Mojave Volatile Prospector field experiment, which simulates a drilling mission at a lunar pole, in October 2014. The ultimate goal is to use the software for an actual lunar prospecting mission that could happen as early as 2018.

This video shows the mission planning tool in action in a simulation of Shackleton Crater, at the Moon’s South Pole. The colored line shows the planned path, with red indicating the robot has plenty of battery power at that point, and blue indicating low battery power. As shadows shift with time, the robot lingers in the sunlight to charge its battery before darting across a shadowed region to reach its goal.

Astrobotic Releases Video on Simulating Reduced Gravity

Gravity offloading is accomplished by applying a force to counter some portion of Earth gravity to achieve an effective reduced gravity. Watch Astrobotic accomplish this with their gravity offload system.

Astrobotic Initiates Testing For Planetary Cave Exploration

Planetary caverns and tunnels can provide shelter from micrometeorites, radiation, and thermal extremes for human and robotic explorers. They may be the best hope for habitation on the Moon. They could be the best place on Mars to find life. They can provide a window into a planet’s past geology, climate, and even biology. Recently discovered skylights, formed by partial cave ceiling collapse, provide access to intriguing but unknown sub-surface voids. “Skylights are gateways to wonders of exploration, science and resources that await beneath planetary surfaces”, said Red Whittaker, Astrobotic CEO. “Robots are our access to those new worlds.”

In a phase I study for the NASA Innovative Advanced Concepts program, Astrobotic developed several mission concepts and investigated key technologies for exploring planetary caves. In phase II, Astrobotic is detailing a mission concept for entering a planetary cave through a skylight, and exploring and modeling the interior using its prototype Tyrobot.

On September 7th 2013, the Astrobotic team conducted a full deployment field test of Tyrobot in an open pit mine in Somerset. The mine served as an analog environment to simulate conditions that Tyrobot might encounter on other planets.


The Tyrobot system uses a trolley to traverse along a cable suspended over the pit. The trolley controls the raising and lowering of a sensor package.


The sensor package has independent power and communications, and logs data from its laser range finder, inertial measurement unit, and fisheye camera.



Those data will be stitched together to create a 3D model of the pit’s interior.

tyrobot 5 tyrobot6

Detailed modeling of planetary pits and caves can reveal geological insights about mineral composition and pit formation. Such models are also valuable for informing subsequent robotic and human missions to explore the intriguing but perilous terrain of pits and caves.

Photos below depict a mission concept to enter and explore a skylight on the Moon using Tyrobot.





Polaris Demonstrates Excavation at NASA Glenn

Polaris is an excavating rover intended to prospect for water ice and other trapped volatiles at the permanently shadowed poles of the Moon. While designed to accommodate a variety of payloads, Polaris is currently fitted with hardware to excavate and transport regolith for mission scenarios involving in-situ resource utilization.

On 5/8/13, an Astrobotic team headed by Chris Skonieczny traveled to NASA Glenn Research Center to demonstrate Polaris’ mobility and excavating ability in lunar regolith simulant. The nominal goal was to dig one metric ton of lunar regolith simulant in one hour. This was the culminating demonstration for a Phase II SBIR contract for Lightweight Robotic Excavation.

Polaris’ testing was conducted at the NASA Glenn SLOPE facility, which provided the means to test mobility and excavating ability in the lunar regolith simulant GRC-1. Polaris met all mobility requirements: reaching a speed of 40 cm/ sec, climbing slopes up to 15 degrees, and traversing a variety of obstacles up to 20cm in height.

Polaris was equally successful at meeting the excavation goals and requirements: digging depths more than 20cm deep, hauling more than 80kg of regolith at one time, and being unaffected by 30cm and smaller rocks. During a continuous excavation test, Polaris excavated, transported, and dumped over 1000kg of lunar simulant in under one hour.

Next for Polaris is to test mobility and excavation performance in a simulated Moon gravity environment. Testing will occur at Astrobotic’s soon to completed gravity offload simulator. While terrestrial excavating machines rely on Earth’s gravity to produce large cutting forces through the soil with a single bucket, Polaris’ continuous excavating system uses multiple cutting buckets that each take a small bite of regolith. This requires lower digging forces, making excavation easier and more productive in low gravity conditions.

EDIT: This article was edited on 10/1/13 to correct an error which listed Kevin Peterson as the lead the for field experiment at the NASA Glenn Research Center.


Follow Astrobotic

Subscribe to me on YouTube

Join Our Mailing List