Abstract
Most evolutionary robotics studies focus on evolving some targeted behavior without taking the energy usage into account. This limits the practical value of such systems because energy efficiency is an important property for real-world autonomous robots. In this paper, we mitigate this problem by extending our simulator with a battery model and taking energy consumption into account during fitness evaluations. Using this system we investigate how energy awareness affects the evolution of robots. Since our system is to evolve morphologies as well as controllers, the main research question is twofold: (i) what is the impact on the morphologies of the evolved robots, and (ii) what is the impact on the behavior of the evolved robots if energy consumption is included in the fitness evaluation? The results show that including the energy consumption in the fitness in a multi-objective fashion (by NSGA-II) reduces the average size of robot bodies while at the same time reducing their speed. However, robots generated without size reduction can achieve speeds comparable to robots from the baseline set.
Introduction
In the field of evolutionary robotics, algorithms are typically employed to automatically design a robot's morphology and controller. However, a significant limitation of many current simulations is the assumption of infinite energy. By ignoring the physical cost of movement, evolution often produces robots with excessive complexity and high-energy "thrashing" gaits that are unsuitable for real-world deployment. This study introduces a battery model into a modular robot simulator to explore how energy constraints act as a regularizer for both physical form and behavioral efficiency.
Methodology
The research utilized the RoboGen framework and Revolve toolkit, with simulations conducted in Gazebo. Robots were composed of modular units, including core components, structural bricks, and active servo joints, with their designs encoded via an L-system grammar.
The primary innovation was the implementation of a battery module that calculates energy drain based on mechanical power (torque multiplied by angular velocity). Using the NSGA-II multi-objective evolutionary algorithm, two experimental setups were compared over 100 generations:
- Baseline: Optimization focused solely on maximizing locomotion speed.
- Battery-Aware: Multi-objective optimization targeting both speed and the maximization of remaining battery charge.
Results and Discussion
The introduction of energy constraints led to several key shifts in the evolved populations. Morphologically, the battery-aware robots were smaller and less complex, averaging 7.44 joints compared to 8.88 in the baseline. This suggests that the battery model effectively penalizes unnecessary structural complexity.
As shown in the figure above, the baseline population achieved significantly higher average speeds. This is largely because the multi-objective approach in the battery-aware experiment maintains a Pareto front that includes highly efficient but slow or stationary robots to preserve battery life. However, a more granular "apples-to-apples" comparison revealed a surprising benefit of energy awareness: when comparing robots with the same number of joints (specifically 9 joints), the energy-aware robots were significantly faster (5.35 cm/s) than the baseline robots (4.33 cm/s). This indicates that energy constraints drive the evolution of more coordinated and effective control strategies rather than simple high-frequency movement.
Conclusion
This research demonstrates that energy awareness is a critical factor in evolving robots that are both physically realistic and behaviorally efficient. By treating energy as a finite resource, evolutionary algorithms can discover high-quality gaits that outperform those found through speed-only optimization for a given body size. Future work will involve testing these energy-aware models in more complex environments and incorporating additional power consumption factors, such as idle electronics drain.
Summary updated