The paper introduces MCD (Multi-Campus Dataset), a comprehensive dataset designed to address various robotics perception challenges. MCD features a wide range of sensing modalities, including classical and non-repetitive epicyclic (NRE) lidars, high-quality IMUs, cameras, and UWB sensors. It covers three Eurasian university campuses, providing diverse environments and domain variations. Key contributions include: 1. **Diverse Sensing Modalities**: MCD includes both classical and NRE lidars, traditional cameras, IMUs, and UWB sensors, offering a rich set of data for perception research. 2. **Semantic Annotations for NRE Lidar**: The dataset provides semantic annotations for 29 classes over 59k sparse NRE lidar scans, addressing the lack of such annotations in existing datasets. 3. **Continuous-Time Ground Truth**: MCD introduces continuous-time ground truth based on optimization-based registration of lidar-inertial data, which is publicly released and significantly larger than existing datasets. 4. **Wider Domain Coverage**: The dataset covers a broader latitude range, providing diverse feature prior distributions and challenging environments. 5. **Challenges in Perception**: MCD tackles real-world challenges such as motion distortion, extreme lighting, and non-line-of-sight observations, making it suitable for end-to-end training scenarios. The paper also includes a detailed statistical analysis of the dataset, benchmarking results on state-of-the-art algorithms, and discussions on the performance and limitations of existing methods. The benchmarks highlight the complexity of MCD and the need for robust and precise solutions in robotics perception.The paper introduces MCD (Multi-Campus Dataset), a comprehensive dataset designed to address various robotics perception challenges. MCD features a wide range of sensing modalities, including classical and non-repetitive epicyclic (NRE) lidars, high-quality IMUs, cameras, and UWB sensors. It covers three Eurasian university campuses, providing diverse environments and domain variations. Key contributions include: 1. **Diverse Sensing Modalities**: MCD includes both classical and NRE lidars, traditional cameras, IMUs, and UWB sensors, offering a rich set of data for perception research. 2. **Semantic Annotations for NRE Lidar**: The dataset provides semantic annotations for 29 classes over 59k sparse NRE lidar scans, addressing the lack of such annotations in existing datasets. 3. **Continuous-Time Ground Truth**: MCD introduces continuous-time ground truth based on optimization-based registration of lidar-inertial data, which is publicly released and significantly larger than existing datasets. 4. **Wider Domain Coverage**: The dataset covers a broader latitude range, providing diverse feature prior distributions and challenging environments. 5. **Challenges in Perception**: MCD tackles real-world challenges such as motion distortion, extreme lighting, and non-line-of-sight observations, making it suitable for end-to-end training scenarios. The paper also includes a detailed statistical analysis of the dataset, benchmarking results on state-of-the-art algorithms, and discussions on the performance and limitations of existing methods. The benchmarks highlight the complexity of MCD and the need for robust and precise solutions in robotics perception.