This a problem of deciding which orders to load before the delivery vehicle embarks? A meter by meter route (of lat/longs) is not provided by Google Maps API and it would be inefficient to check proximity to a waypoint every meter anyhow. You could pull a set of two waypoints from the XML/JSON return dataset and create a rectangle around it and see what lat/longs (for your undecided orders) fall inside the rectangle, but some roads curve or are semi-circular. The best solution is actually to display the orders on a map and allow a human to draw a circle around the orders (thus selecting them) and then calculate a route and have a human adjust it once more (route optimization algorithms do not derive the most efficient route). Humans know which side of town can be grouped to together (and at what time of day) far more efficiently than an algorithm will.
I cannot give you a complete bid to complete the job because of my freelancer status limitations. But I can tell you that completing the job will cost more than what you are offering. I can offer you design work at this price. the price to build would have to be quoted separately.
I would include a writeup of the work I did for FTD, a delivery desktop application for a point-of-sale system, but it exceeds the character limit for the proposal.