Difference between revisions of "Energy Optimization"

Revision as of 07:57, April 12, 2013

Energy optimization is a fundamental requirement to achieve long term autonomous deployments of mobile robots. Most robotic systems which operate on the field are subject to severe energy limitations. In particular, robot developers face trade-offs between payload capacity and battery life since existing batteries are heavy and last a short amount of time under full actuation. Hence, it is critical to develop energy-aware algorithms that optimize the existing resources on the robot and harvest additional energy when available.

We frequently encounter this problem of limited on-board energy in our field experiments for the Data Mules and Environmental Monitoring projects. With a goal of achieving long-term autonomy, we are studying the following problems.

Trajectory Planning to Minimize Energy Consumption

One of the main bottlenecks for robots is the limited lifetime of on-board batteries. Motion is a major source of energy consumption. In [7], we studied the problem of minimizing the energy consumption for a car-like robot by optimizing its path and velocity profile. We make the following contributions:

• When the path for the robot is given as input, we present a closed-form solution for finding the optimal velocity profiles for the robot that minimize the energy consumption (see [2])
• For finding minimum energy paths, there is a trade-off between the length of the paths, turning radius, frictional forces, velocity and acceleration of the robot. Unlike minimum length (Dubins') paths, a minimum energy path may contain segments with varying turning radius. We show how to efficiently construct a graph which generalizes Dubins’ paths by including segments with arbitrary radii.
• We present an algorithm uses that the closed-form solution for the optimal velocity profiles as a subroutine to find the minimum energy trajectories, up to a fine discretization. We investigate the structure of energy-optimal paths and highlight instances where these paths deviate from the minimum length Dubins’ curves.
• Finally, we present a calibration procedure to obtain the energy model and validate our algorithms with experiments conducted on a custom-built robot.

A software package that implements the above contributions is available for download here.

Energy Harvesting

In addition to optimizing the onboard energy consumption, we are studying how to efficiently harvest solar energy on robots. See Solar Power project page for details.

Related Publications