I think about how to improve the RoboComp system by adding visual detection mechanisms to the mobile robots. This provides to the robot all the information needed to predict its short-term environment and react better than a robot without this technology (it could look less human, it couldn't know about its probable near future so it couldn't react to it). That is to say, if a person passes in front of the robot, probably, it can't know it to react and border the person, because it is an unpredictable event. But, with this technology, if a human crosses ahead, the robot detects a change in its environment and it can change its path to not collide with the human. I have used RoboComp before and I know that it is a good software that can be greatly improve with this system.