Monocular Visual Odometry for Mars Exploration Rovers

Over the past twenty years Mars surface exploration has been conducted particularly by the following six-wheel rocker-bogie mobile rovers developed at NASA Jet Propulsion Laboratory [1]: the Mars Pathfinder mission rover Sojourner, the Mars Exploration Rovers (MER) Spirit and Opportunity and the Mars Science Laboratory (MSL) rover Curiosity, where the latter is still on his way to Mars and is expected to land on Mars on August 2012 [2]. Because the maximum exploration range of a rover is limited to few tens of kilometers, a Mars Airplane is also currently being developed [3], which will increase the mission coverage to hundreds of Kilometers. As opposite to the Mars Airplane, flapping insect robots (entomopters) are also being investigated due to their potential to fly slow as well as safety land and take off on the rocky Mars terrain [4]. Since the rovers are typically commanded only once per Martian solar day, they must be able to autonomously navigate to science targets and to place instruments precisely against these targets, where any navigation error could cause the loss of the entire day of scientific activity. For precise autonomous navigation, the rovers must have an onboard system for precise and reliable estimation of its position and orientation. Usually the current rover's position and orientation are estimated by integrating the rover's motion (rover's change of position and orientation) from the time the motion began to the current time, assuming that the initial rover's position and orientation are known or previously estimated. In the MER rovers Spirit and Opportunity the rover's change of orientation (rover's rotation) is estimated from measurements of three-axis angular rate sensors (gyros) provided by an Inertial Measurement Unit (IMU) onboard the rover [5]. The rover's change of position (rover's translation) is estimated from encoder readings of how much the wheels turned (wheel odometry). The initial rover's orientation is estimated from measurements of three-axis accelerometers provided by the IMU, as well as a sun position vector provided by a sun sensor which is also onboard the rover. The initial position is reset by command at the beginning of the rover's motion. Unfortunately, a limitation to the rocker-bogie mobile rovers as observed on Mars is excessive wheel slippage on steep slopes, which causes large errors particularly on the estimated rover's position from wheel odometry. To correct any position error, the rover's motion is also estimated by using a stereo visual odometry algorithm, which estimates the rover's motion by maximizing the conditional probability of the 3D correspondences between two sets of 3D feature point positions, which were previously obtained from two consecutive stereo image pairs captured by a stereo video camera before and after the rover ́s motion, respectively. The conditional probability is computed by modeling the 3D position error at each feature point with Gaussian distributions and using a linearized 3D feature point position transformation, which transforms the 3D position of a feature point before motion into its 3D position after motion given the rover's motion parameters. This stereo visual odometry algorithm was first proposed by Moravec in [6] and then improved in [7][8]. Afterwards, it evolved to become more robust [9] until it was finally implemented in real time to be used in the Mars Exploration Rover Mission [10]. After evaluating the performance of the above stereo visual odometry algorithm in both MER rovers Spirit and Opportuniy on Mars, it was further improved in [11] resulting in a more robust and at least four time more computationally efficient algorithm, which can also operate with no initial motion estimate from wheel odometry. This last updated version of the stereo visual odometry algorithm is planned to be used in the MSL rover Curiosity. Our idea is to develop, implement and test in real time and in a real rover test-bed a monocular visual odometry algorithm which will be able to estimate the robot's motion evaluating the intensity differences at different observation points between two intensity frames captured by a monocular video camera before and after the robot's motion, respectively, as an alternative to the traditional stereo visual odometry. The rover ́s motion will be estimated by maximizing the conditional prob4053.pdf Concepts and Approaches for Mars Exploration (2012)