Collaborative robots (Cobots) are increasingly being used as assistance systems to relieve humans in private and commercial environments. Current systems are designed to stop in the event of a collision, which means that robots only stop after unwanted contact. For completely safe physical collaboration, collisions should be avoided as far as practicable, and distance measurement should take place instead. The internal sensors of Cobots typically do not provide exteroceptive distance measurement, which is why additional sensors must be added. To ensure safe collision avoidance, low-latency systems are required to monitor the work area and avoid collisions. This work investigates how different integration methods for distance-sensors in ROS2 affect end-to-end latency when running on a resource-constrained microcontroller. A Raspberry Pi Pico is used to connect two low-cost sensors that are representative of typical HRC cells: an ultrasonic-sensor and a time-of-flight-sensor. Both sensors use different communication standards. Two integration strategies are compared. In the standard ROS2 integration, the sensor firmware runs on the microcontroller, which streams the data via a serial connection to a ROS2 node. In the micro-ROS-based integration, the firmware and ROS2 publisher are combined in a single application. For each sensor, the latency between measurement acquisition and publication in a ROS2 topic is quantified over multiple trials. The results show that micro-ROS reduces average latency and suppresses extreme outliers, although the overall update rate remains limited by the sensor’s intrinsic sampling frequency. The study derives practical design guidelines for latency-aware integration in ROS2 on low-cost microcontrollers and discusses their implications for safe HRC.