Layer 01
Sensors
Camera, IMU, semantic cues, communication signals, and future event-based sensing.
Modules
CameraIMUEvent streamChannel state
Connected projects
Adaptive Multi-Modal SLAMOpenUWOC-AINeuromorphic Robot SNN
System architecture
A robotics autonomy stack, not a project list.
This architecture explains how the research modules connect from sensing to perception, state estimation, uncertainty, risk, planning, navigation, safety, and simulation. It is conceptual and honest: planned modules remain labelled as future work until implemented and evaluated.
Layer 02
Perception
Scene understanding that supports navigation rather than only producing visual labels.
Modules
SegmentationTraversabilityFailure casesReliability
Connected projects
Urban Segmentation Delivery RobotReal-Time Driving Scene SegmentationSafeCrossAI
Layer 03
State Estimation
Localization and mapping under uncertainty, degradation, and sensor-fusion constraints.
Modules
VIOSLAMSensor fusionEstimator health
Connected projects
SHIELD-VIOAEGIS-VIOAdaptive Multi-Modal SLAM
Layer 04
Uncertainty + Risk
Explicit representation of confidence, degradation, risk exposure, and returnability.
Modules
Health indexRisk mapsPrediction uncertaintyReturnability
Connected projects
DynNavSafeCrossAIRisk-Aware UAV RTH
Layer 05
Planning + Navigation
Decision-making modules that trade off efficiency, safety, uncertainty exposure, and recovery.
Modules
ReroutingSafe modeRTH policyPlanner comparison
Connected projects
DynNavuncertainty-aware-navigationRisk-Aware UAV RTH
Layer 06
Simulation + Evaluation
Synthetic but clearly labelled experiments for reproducible development before validated benchmarks.
Modules
Seeded scenariosMetricsMission logsPending results
Connected projects
Formula 1 Race SimulationDynNavSHIELD-VIO
Architecture rule
The diagram is a research operating model, not a claim of a deployed full robot stack. Each block points to current, prototype, planned, or draft work. Quantitative performance remains pending until reproducible results exist.