This paper presents the design and performance evaluation of a low-cost Attitude and Heading Reference System (AHRS) for autonomous vehicles. A single sensor pack, an Inertial Measurement Unit (IMU), provides all the data required to feed the attitude filter. The design is sensor-driven and departs from traditional solutions as no explicit representations of the attitude, e.g., Euler angles, quaternions, or rotation matrices, are considered in the filter design. Moreover, the proposed solution includes the estimation of rate gyros biases, systematic tuning procedures, and also allows for the inclusion of frequency weights to model colored noise on the different sensor channels. Due to its inherent structure, the filter is complementary, allows for temporary loss of sensor measurements, and also copes well with slowly time-varying rate gyros biases. The performance of the proposed algorithm is experimentally evaluated with a low-cost IMU and resorting to a high precision calibration table, which provides ground truth signals for comparison with the resulting filter estimates.