<p>This paper contributes to the development of an online learning and adaptive control law for networked 2-degree-of-freedom (2-DOF) laboratory helicopters, utilizing smartphones and embedded computers. The primary objective is to address the challenges associated with controlling motion trajectories, specifically time-varying pitch and yaw angles, of fixed-base helicopters with partially known dynamics. These challenges are further amplified when coordinating multiple helicopters simultaneously using onboard embedded computer and a smartphone application. The proposed methodology employs reinforcement learning (RL) formulated as an approximate dynamic programming (ADP) problem to design an adaptive control strategy. Unlike conventional control approaches, such as linear quadratic regulators (LQR), H-infinity control, or proportional–integral–derivative control, that often assume complete knowledge of system dynamics or are validated in simulations or single-helicopter setups, the ADP technique introduced in this work leverages real-time state measurements to compute actuator commands. This enables effective tracking of reference motion trajectories without relying on full system dynamics. Additionally, the ADP framework is augmented by implementing an optimal output tracking and a linear quadratic regulator controller as benchmarks for comparative analysis. Experimental results demonstrate the efficacy of the proposed RL-based ADP control strategy in achieving accurate trajectory tracking for multiple networked helicopters in real-time. The performance comparison highlights significant improvements in adaptability and robustness over the benchmark controller. This work contributes to the field of intelligent control by demonstrating a practical and scalable approach to real-time motion trajectory control for networked multi-agent helicopter systems, validated through embedded computing and smartphone integration.</p>