Battery aware reinforcement learning improves aggressive drone flight control
Battery-Aware Reinforcement Learning for Aggressive Quadrotor Flight
Robotics
Summary
Flying drones fast and precisely is hard because their motors can’t always push as hard when the battery runs low. The authors show how teaching a drone's controller to pay attention to battery voltage during training and flight helps it use more power safely, making turns and circles more accurate. Testing on a small drone, this approach made flights faster and more precise, especially in challenging conditions like racing. They also show that giving the controller the right battery voltage information is important for the best performance.
What this means in practice
- •For drone developers: Design drone flight controllers that adapt thrust commands based on current battery voltage for better race and maneuver performance.
- •For robotics engineers: Incorporate battery state input into learned control policies to improve precision in aggressive quadrotor tasks like pursuit or inspection.
Authors
Alejandro Sanchez Roncero, Olov Andersson, Petter Ogren
Abstract
Agile flight tasks such as drone racing and pursuit-evasion require strong acceleration and precise turns, but the available thrust changes as the battery discharges and voltage drops under load. Conservative command limits make this variation easier to tolerate, at the cost of unused performance. We investigate how learned controllers can use that additional thrust while retaining the flight controller's voltage compensation and rate control. Our training simulator couples an identified load-transient battery model to rotor dynamics and firmware saturation. The feedforward policy receives filtered voltage during both training and deployment. Controlled ablations distinguish the benefit of a larger thrust-command range from that of voltage information. On a 38 g Crazyflie Brushless, the resulting policy reduces circle tracking error by 49% relative to stock-authority RL at 3.84 m/s, while preserving easy-task precision. Mean 20-lap race time decreases from 106.22 s to 95.24 s. Compared with a voltage-blind policy with the same increased authority, hardware error and race time are lower by 15.3% and 4.5%, respectively. In simulation, replacing the policy's voltage input with a recording from a different battery condition worsens hard-circle tracking, with a smaller, voltage-dependent effect in racing. Together, these results show where a simple voltage input complements existing actuator compensation in aggressive learned flight.