Problem with the demo.

[Gatsby23]

Thank you for your contributions to the research; your work has significantly propelled advancements in the area of multi-map merging. However, I have encountered some technical challenges while recently testing the ROVIOLI algorithm and would appreciate your guidance. My system runs on Ubuntu 20.04, equipped with an Intel i9 processor and an NVIDIA RTX 4090 graphics card. Based on recommendations from the issue380 I executed the following commands:

roslaunch maplab_node euroc-maplab-node-w-rovioli.launch 

The test results indicate there is a memory leak, with the specific details as follows:

*** Aborted at 1737997072 (unix time) try "date -d @1737997072" if you are using GNU date ***
PC: @                0x0 (unknown)
*** SIGSEGV (@0x200000002) received by PID 2996886 (TID 0x7fccee7ec000) from PID 2; stack trace: ***
    @     0x7fcda5ace420 (unknown)
    @     0x7fcda42f4a5a cv::ellipse()
    @     0x7fcda03c9cb4 rovio::FeatureCoordinates::drawPoint()
    @     0x7fcda581c679 rovio::ImgUpdate<>::extraOutlierCheck()
    @     0x7fcda5842459 LWF::Update<>::performUpdateIEKF()
    @     0x7fcda5845a6b LWF::FilterBase<>::update()
    @     0x7fcda5846bb9 LWF::FilterBase<>::updateSafe()
    @     0x7fcda573aa41 rovio::RovioInterfaceImpl<>::updateFilter()
    @     0x7fcda5847798 rovio::RovioInterfaceImpl<>::processImuUpdate()
    @     0x7fcda589a0dd _ZNSt17_Function_handlerIFvRKSt10shared_ptrIN3vio14ImuMeasurementEEEZN7rovioli9RovioFlow19attachToMessageFlowEPN12message_flow11MessageFlowEEUlRKS0_IKS2_EE_E9_M_invokeERKSt9_Any_dataS5_
    @     0x7fcda58a28a0 message_flow::MessageDeliveryQueue<>::deliverOldestMessage()
    @     0x55d38470e565 _ZNSt17_Function_handlerIFSt10unique_ptrINSt13__future_base12_Result_baseENS2_8_DeleterEEvENS1_12_Task_setterIS0_INS1_7_ResultIvEES3_EZNS1_11_Task_stateISt5_BindIFSB_IFMN12message_flow24MessageDeliveryQueueBaseEFvvEPSD_EEvEESaIiEFvvEE6_M_runEvEUlvE_vEEE9_M_invokeERKSt9_Any_data
    @     0x55d38470f02d std::__future_base::_State_baseV2::_M_do_set()
    @     0x7fcda5acb4df __pthread_once_slow
    @     0x55d38470f331 _ZNSt17_Function_handlerIFvvEZN5aslam10ThreadPool14enqueueOrderedISt5_BindIFMN12message_flow24MessageDeliveryQueueBaseEFvvEPS6_EEJEEESt6futureINSt9result_ofIFT_DpT0_EE4typeEEmOSE_DpOSF_EUlvE_E9_M_invokeERKSt9_Any_data
    @     0x7fcda5a32382 aslam::ThreadPool::run()
    @     0x7fcda53aa793 (unknown)
    @     0x7fcda5ac2609 start_thread
    @     0x7fcda507d353 clone

My gcc and g++ versions are both 9.4.0. Considering the issue might occur under the VIO mode of ROVIOLI, I also independently tested the ROVIO algorithm, which can reproduce the expected results on the EuRoC dataset. How should I address this bug? Additionally, is it possible to directly replace the maplab_rovio codebase with the ROVIO codebase?