Proactive Automatic Gear Shift Proposal

Problem

Automatic gear shifting in vehicles today is less efficient than manual transmissions. This is due to a multitude of problems such as slippage, more internal friction, and less optimal gear selection. Automatic transmissions select gears through a reactive process. They use factors like vehicle speed, throttle position, and engine load to make a guess about what gear is correct. We believe that gear shifting can be done more efficiently if a proactive approach is taken where the current and incoming terrain are taken into consideration.

Project Outline

In order to address this problem, we will develop a machine learning model that determines the optimal gear to shift to based on current and incoming terrain, resulting in a proactive approach. The model will use video and/or lidar information coupled with data about the available motors (human or non-human) to determine optimal gearing. 

This involves developing a data-collection method for terrain, available power, and gearing, possibly including the production of hardware or the development of an iPhone application. Next steps include training and experimentally verifying the model. Extensions include developing a feed-back loop which allows the model to learn from successful and unsuccessful shifts and the development of a product/implementation using the model.

Applicability

More efficient gear selection for transportation would have a myriad of applications. For both manned and unmanned vehicles, energy efficiency is incredibly important, especially as energy prices are rising. This technology could be applied to all vehicles from bikes, to cars, to automated off-road vehicles.