The purpose of this project is to develop a framework containing two subsystems to aid in indoor UAV flight. The two systems will be developed and tested independently. A visual odometry system to be able to estimate motion in GPS denied areas, and an obstacle avoidance system to help avoid collisions. For the visual odometry system a novel method using a downwards and upwards facing camera is proposed. Features are tracked using Shi-Tomasi and Lucas-Kanade. From the tracked features a homography is estimated for each of the cameras. These are combined into a single motion estimate. For the obstacle avoidance system, a method is proposed which divides the image into left, right and center region and classifies these independent. Features are tracked using Shi-Tomasi and Lucas-Kanade, the tracked features are combined into trajectories. The trajectories are classified using the proposed method. The visual odometry system is found to have a mean absolute error varying from 0.16cm up to 0.54cm per sample. Errors in rotation have been found to vary from 0.13º up to 0.19º. For the obstacle system 82 different tests have been carried out in 3 different scenarios. The left and right region is classified with a 78.3% detection rate. The center region is classified with a 100% detection rate.