LACART (Lookahead‑Based Autonomous Racing & Control Toolkit) merges a lookahead control‑barrier function, a relaxed control‑Lyapunov function, and a single quadratic program to adjust a robot’s nominal drive commands. The quadratic program balances error from the desired trajectory with safety constraints using an input weight matrix **W** and a relaxation penalty **p_s**. In simulation, LACART achieves a median per‑step computation time of 0.31 ms, about 6.1 % faster than the baseline method of reference [9] (0.33 ms). The median quadratic‑program solve time drops from 0.025 ms to 0.023 ms, an 8.0 % reduction. These speed gains come with no loss of tracking accuracy.

Safety proofs require the lookahead point to remain a safe distance from obstacles and track boundaries. For a disk‑shaped robot, the minimum clearance **d_min** must at least equal the lookahead distance **l**, and when a bounded position error **η** exists, the condition becomes **d_min ≥ l + η**. State‑estimation latency (average 62.44 ms) can translate the robot by up to 6.24 cm at 1 m/s, which shrinks the effective safety margin.

Hardware tests confirm that the full control loop runs every 40 ms, with the QP taking roughly 7.5 ms—about 19 % of the loop period—while image acquisition, image processing, and communication occupy the remaining time. These experiments also revealed occasional collisions with track boundaries when steering limits and delays are hit, and a brief halt when both side boundaries activate simultaneously. Such results underline the need to account for real‑world delays and actuator limits.

Overall, LACART demonstrates that integrating lookahead CBFs with relaxed CLFs in a single quadratic program can provide fast, provably safe control for autonomous racing robots while highlighting key practical limitations like latency and hardware constraints.