In this article, we explore a novel approach for the extrinsic misalignment calibration of automotive radar sensors. In a previous publication we introduced a method for online estimation of the side-slip gradient using radar sensors. We now propose using this estimated side-slip gradient to improve radar misalignment calibration. Current literature usually neglects the lateral velocity at the vehicle's rear axis, when calibrating the radar misalignment. This limits the achievable accuracy and availability of the method. Thus, given our previous success in estimating the side-slip gradient online; we propose an algorithm that estimates both the vehicle ego-motion states and the misalignment of multiple radar sensors, along with the side-slip gradient. Our findings demonstrate that this integrated approach improves the accuracy of radar misalignment calibration, while also providing an accurate estimate of the side-slip gradient. Empirical validation using multiple data sets shows that the algorithm is robust to changing conditions and provides consistent accuracy. Thus, this approach can be applied to series-production vehicles.