The research into extrinsic calibration of LiDAR with a stereo rig has been relatively dormant. Meanwhile, the sensor combination has seen a surge in popularity due to their strengths in usage on autonomous vehicles. In that application, high accuracy calibration can be of paramount importance when an autonomous vehicle takes choices of centimeter accuracy based indirectly on this calibration. For the sake of robustness, if e.g. an incident occurred, upsetting this calibration, one would not want to send the vehicle back to the manufacturer for recalibration by an expert. One would rather hope to automatically run an in-field recalibration routine and then continue operations as normal. This thesis proposes an approach to calibration of the aforementioned sensors by direct point cloud alignment, using the Iterative Closest Point (ICP) algorithm, allowing for the aforementioned in-field recalibration capacity. Included in the approach is a validation scheme, aimed at detecting stable convergence, as well as providing error bounds for the found calibration. In theory our approach is also applicable to any sensor combination, able to produce point cloud data, making its usability broader than at first glance. The method proposed has been thoroughly tested on real world data from an Unmanned Ground Vehicle (UGV), showing some promise, easily able to outperform any manual scheme for aligning the sensor data, through measurements or otherwise. One of the main issues facing data-driven calibration approaches, local minima in the optimization landscape, has been shown to also be the main issue facing our approach, as the discreetness of data leads to oscillations in the optimization landscape.