Risk Analysis

S No.DescriptionTypeLikelihoodConsequenceRisk Management Evaluation
1.
Slow model execution during simulationT24Success: Only bottleneck is data throughput of the
CARLA simulator. Our system theoretically can
provide 100 FPS.
2.
Performance of learning
model is not adequate		T34Success:
Fallback rule based model was implemented.
3.
Inability to formulate
the modelling problem		S,M22Success: Since we had already prototyped a model
early in the spring, we managed to avoid this risk
4. Missing project milestonesS, M25Success:
- We were able to achieve all the required requirements.
Failure:
- We were not able to achieve all the desired requirements
5Stakeholder disengagement M32Success: We always had regular meetings with our sponsors
and discussed all findings with them and we received
full and continued support from them throughout
6Inadvertent failure during demosS, M24Success: Our risk reduction strategy was successful here and
we did not face this risk during the demos.
7Unable to capture data due to
COVID-19 pandemic extension		S, M42Success: We were able to capture data outdoors as
the COVID situation was under control at the time
8Insufficient compute power
for training models		T44Success: We procured a powerful machine early on in
the project
9Unable to procure
video capture license		S, M42Success: We got approval from IRB for capturing data.
10Lead time of hardware
causes delays in
development/impacts
the schedule		S33Success: Our choice of hardware was readily
available online.
11Detection and tracking
algorithms don’t work
in different weather conditions		S, T42Success:
- We were able to process data with varying weather in
simulation.
Failure:
- We removed the varying weather conditions from the real
world as we were not able to capture and annotate
weather data
12Lead time of hardware
causes delays in
development/impacts the schedule		S33Success: We ordered our components well in advance