Heterogeneous multi-robot Simultaneous Localization and Mapping (SLAM) is the problem of building a map and localizing in it with multiple agents using different kinds of sensors, such as Lidars (light detection and ranging) and stereo cameras. While many solutions exist for both associating information from sensors of different kinds, and for creating a map with multiple sensors of the same type (homogeneous sensors), research in the combined topic is still at an early stage. At the same time, the amount of devices with sensors is increasing by the day, and providing a method for these to collaboratively map and localize in environments would decrease the need for device-specific solutions. This thesis proposes a heterogeneous multi-robot SLAM framework based on SegMap as a common map representation for Lidar and stereo sensors. While a SLAM framework based on SegMap already exists, it was originally designed for Lidar sensors only. The proposed method therefore explores which aspects that need to be addressed in the transition from homogeneous to heterogeneous SLAM with multiple sensors. First a baseline framework is benchmarked, and then general extensions are proposed that could help improve performance in heterogeneous scenarios. These extensions are implemented and tested, giving new results that highlight in which cases the extensions were useful, and point towards the next steps in heterogeneous multi-robot SLAM. ; Heterogen SLAM (eng. simultaneous localization and mapping) med flera robotar är ett problem inom robotik som handlar om att flera enheter med olika sensorer tillsammans kartlägger ett område och samtidigt lokaliserar sig i förhållande till kartan. Flera lösningar finns både för att associera information från olika sensorer, såsom Lidar (eng. light detection and ranging) och stereokamera, och för att kartlägga områden med flera likadana sensorer (homogena sensorer), men det kombinerade ämnet är fortfarande relativt nytt. Samtidigt ökar också mängden enheter med sensorer i samhället, vilket ...