Social Motion Planner Implementation

Warning: Shameless plea for help ahead :)

Hi, I am graduate student doing a project on pedestrian collision avoidance. I am currently looking for relevant matlab submissions, libraries, and/or toolboxes which would help me develop the motion planner for a mobile robotic system (autonomous). The problem is I have a month to get the system up an running so I need to develop a quick and easy implementation which can work with C++/java/python so I can send data directly from a motion planner in matlab via a raspberry pi to a controller of an RC car (see attached picture). I have heavily constrained the system to have only a few obstacles and am using a 2-wheel RC Car. My Requirements are that there is a single pedestrian and stationary obstacles which the car has to avoid, a little (or big problem) I am currently trying to solve is that the RC Car cannot support any sensors on it so I have to use a computer vision system to track both obstacles (including pedestrian) and the RC Car. Localization will be done entirely through the web-cam and the speed of the RC-Car will be adjusted to match the speed of the localization (it just has to work no matter how slow). The vision system will be tracking the car using colored ping pong balls (possibly with LEDS inside) and using a multi-hypothesis tracking localization approach (I haven't chosen a specific algorithm) with SIFT for tracking the ping pong balls.

So to recap I need help with the following things:

• Easily Modifiable Motion Planner which can be augmented to approaches shown in the attached papers.
• Bi-directional real-time communication between matlab and a raspberry pi for autonomous control.
• Any information, links, examples of autonomous controllers for small 2-wheeled robots.
• Any advice in general.

Thanks for your help in advance, and sorry for the wording (its late at night) :)!