When deploying autonomous systems that require several sensors for perception, accurate and reliable extrinsic calibration is required. In this research, we offer a reliable technique that can extrinsically calibrate numerous lidars in the base frame of a moving vehicle without the use of odometry estimation or fiducial markers. Our method is based on comparing the raw IMU signals between a collocated IMU present with the lidar and the IMU measurements from the GNSS system in the vehicle base frame. Additionally, based on our observability criterion, we choose measurements that include the most mutual information rather than comparing all comparable IMU readings. This enables us to locate the measurements that are most useful for real-time calibration. Utilizing data gathered from Scania test vehicles with various sensor setups, we have successfully validated our methodology.
翻译:当部署需要多个感知传感器、准确和可靠的外部校准的自动系统时,需要使用几个感知传感器、准确和可靠的外部校准。在这项研究中,我们提供了一种可靠的技术,可以在不使用奥多里亚估计或浮标的情况下,在移动车辆的基底架上对许多利达进行外部校准。我们采用的方法是比较在车辆基架上带利达的同一地点的IMU和从全球导航卫星系统系统中测量的IMU之间的原始IMU信号。此外,根据我们的可观察性标准,我们选择的测量方法包括最相互的信息,而不是比较所有可比较的IMU读数。这使我们能够找到对实时校准最有用的测量方法。我们利用具有各种感应装置的Scania测试车辆收集的数据,成功地验证了我们的方法。</s>