This paper presents an approach to enhance the perception quality of a multi-sensor system by considering the context information. The platform's context state is obtained by combining data from the perception sensors and from additional context sensors. This context information is then utilized to dynamically adapt the sensing/processing parameters to the current context. Additionally, the system is capable to detect a reduced perception performance of individual sensors caused by environmental influences.The proposed approach was implemented on a multi-sensor platform, equipped with Time-of-Flight (ToF) cameras, radar sensors and multiple context sensors. The static Robot Operating System (ROS) architecture of the existing platform was extended to support automatic parameter adaption during runtime. In order to demonstrate the approach with real-world data, the platform was exposed to different scenarios. The proposed approach results in a significantly increased perception quality compared to the output of a static implementation.