On-board collection and storage of road data and implementation of predictive transmission fuel-saving functions

Abstract

Fuel-saving functions are of great concern to truck makers today. A desired function for achieving this is the ability to access preview data about the upcoming road. This feature can be useful for a lot of purposes in many areas. Beside fuel-savings, one can for example achieve a better driving comfort and a more intelligent cruise control. Preview data can be read from a digital map, but since digital road maps containing height information are commercially rare and do not cover all roads, this thesis has looked at another solution, namely to develop a self-learning system to provide the truck with road information without an initial database.

This thesis has the focus to develop a method to create a digital road map independent of whether road maps exist or not. This is done by using the truck's built-in inclination sensor and a GPS for positioning. Since the available memory is highly limited in existing ECU's, only the road where the developed functions have the biggest effect are saved and hence partial maps are created.

To test and evaluate the developed system, a previously developed function that uses preview data has been implemented. The function is active shortly before crests and in the following downhill slope and can save approximately 1.5 percent of fuel in addition to existing fuel-saving functions.

Since the memory is limited, it has been investigated how to represent a road with respect to minimising the data needed to be stored. The simulation results show that it is sufficient to use 8 points to characterise a hill, and then perform a clamped spline interpolation for restoring the data to create a functional preview. With this coding of height information, the assigned amount of memory allows information about up to 1500 hills or about 5000 km of road to be stored. The self-learning algorithm can be implemented in existing hardware without any additional equipment.