Evaluating Robot Program Performance with Power Consumption–Driven Metrics in Lightweight Industrial Robots

The code performance of industrial robots is typically analyzed through CPU metrics, which overlook the physical impact of code on robot behavior. This study introduces a novel framework for assessing robot program performance from an embodiment perspective by analyzing the robot’s electrical power profile. Our approach diverges from conventional CPU-based evaluations and instead leverages a suite of normalized metrics, namely, the energy utilization coefficient (f<inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">U</inf>), the energy conversion metric (f<inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">C</inf>), and the reliability coefficient (f<inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">R</inf>), to capture how efficiently and reliably energy is used during task execution. Complementing these metrics, the established robot wear metric (α) provides further insight into long-term reliability. Our approach is demonstrated through an experimental case study in machine tending, comparing four programs with diverse strategies using a UR5e robot. The proposed metrics directly compare and categorize different robot programs, regardless of the specific task, by linking code performance to its physical manifestation through power consumption patterns. Our results reveal the strengths and weaknesses of each strategy, offering actionable insights for optimizing robot programming practices. Enhancing energy efficiency and reliability through this embodiment-centric approach not only improves individual robot performance but also supports broader industrial objectives such as sustainable manufacturing and cost reduction.