Autonomous space vehicle navigation
This work done at the Autonomous Systems Lab under the guidance of professor Kartik B. Ariyur looks at improved methodology to estimate the state of a space vehicle in order to increase the autonomy and reduce the cost. This method is based on a multi-sensors approach of low SWAP (size, weight and power) imagers. This method reduces the cost on two levels. Low SWAP sensing units reduces the cost of the spacecraft. Increased autonomy reduces the operational cost (house- keeping, on-orbit correction). The sensors used are low cost imagers in order to avoid drawback from Inertial Measurement Unit (dead reckoning issue) or other inertial instruments and to avoid power consuming sensing units. Also, the cameras allow us to work with absolute measurements. The methodology is based on multiple cameras on each sides of the spacecraft. Each camera should recognize a minimum of 5 stars to solve the system of equations however a bigger number of stars is required to achieve a better accuracy. One of the key steps in solving the problem is to reduce the non-linear into a linear system. Then solving separately for the attitude and the position. The variables of the heavily overdetermined linear system can be estimated by using a constrained least-square method. The low SWAP sensors used for the simulation are the Samsung S5K6A3 CMOS imagers in order to validate the proposed method. The reference trajectory is a low-Earth orbit at an altitude of 120 km. The spacecraft is constantly spinning along a single axis. The Monte Carlo campaigns determine the maximum error boundary for two different et of stars: 150 stars and 1000 stars. We can clearly observe that the 150 stars configuration is subject to delay before convergence and reduction of the error. On the other hand, the simulation with 1000 stars avoid this convergence issue and provides high accuracy in the attitude estimation (error of less than 0.4 deg). This work proves the feasibility of the proposed method with relatively high ac- curacy meeting requirements of most scenarios and space applications. This allows significant cost reduction for commercial applications.
Ariyur, Purdue University.
Off-Campus Purdue Users:
To access this dissertation, please log in to our