01 / LiDAR
High-resolution geometry
2D laser scanning provides detailed range and obstacle geometry in clear conditions, but may suffer from scattering and backscatter in fog.
SPATIAL PRECISION UCSB · ECE 189 CAPSTONE PROJECT
See beyond
the visible.
Autonomous navigation doesn't stop when visibility does. We're building a rover that combines the precision of LiDAR with the resilience of millimeter-wave radar.
RADAR–LIDAR FUSION ROVER DEGRADED-VISIBILITY NAVIGATION
PERCEPTION / CONCEPT VIEW SENSOR FUSION
360° LiDAR coverage60 GHz radar bandILLUSTRATION ONLY
02Complementary sensors
03Sensing modes to compare
05Engineering team members
1 m/sTarget maximum rover speed
01 / THE CHALLENGE
When the environment gets uncertain,
perception matters more.
Fog, smoke, and dust can scatter laser light and degrade optical sensing. Radar offers a different perspective, but its measurements can be sparse or noisy. EchoVision investigates how both can work together for safer autonomous navigation.
02 / mmWave radar
Resilient sensing
Millimeter-wave measurements help detect objects when optical visibility is compromised, with different resolution and clutter trade-offs.
ROBUST DETECTION03 / Fused perception
Complementary strengths
Time-aligned measurements and filtering aim to make obstacle detection and navigation more reliable across changing environments.
OUR RESEARCH QUESTION02 / SYSTEM DESIGN
Built as one
connected system.
A modular robotics architecture separates perception and planning from safety-critical low-level control. Component selections remain subject to integration testing.
01 / SENSE
2D LiDARRPLIDAR S2 candidate
mmWave RadarTI IWR6843ISK candidate
→
02 / PERCEIVE + PLAN
Main Computer + ROS 2Sensor processing · timestamp alignment · fusion · obstacle detection · navigationRaspberry Pi 5 proposed for initial development; Jetson considered as an upgrade
→
03 / CONTROL
STM32 MCUIMU · control logic · communication
Motor Driver + RoverMotion · speed · emergency stop
A
Observe
Acquire range, target, and motion measurements across sensors.
B
Decide
Filter detections, estimate obstacles, and determine safe navigation actions.
C
Act
Translate decisions into motor commands with dedicated low-level control.
03 / EXPERIMENTAL PLAN
Measure the difference.
Not just the promise.
Our planned evaluation compares three sensing configurations under consistent test conditions. Results will be published as the rover is built and validated.
CONFIGURATIONPLANNED EVALUATION
LiDAR only
Baseline geometry and detection performance in clear and degraded visibility.
Radar only
Baseline target detection and clutter behavior under the same conditions.
Radar + LiDAR fusion
Assess whether combining both sensors improves reliable detection and stopping response.
Detection rangeHow far obstacles can be reliably detected
False alarmsHow often non-obstacles trigger detections
Stopping latencyTime between detection and safe response
Battery runtimeDuration of operation on onboard power
CONTROLLED TEST ENVIRONMENT
Clear air → light fog → dense fog
Planned controlled fog trials use an enclosed test area, an optical fog-opacity measurement, and representative obstacles such as boxes, thin poles, transparent barriers, and moving targets.
EXPERIMENTS IN PLANNING ↗
04 / THE PEOPLE
Five people.
One shared vision.
EchoVision is a five-member engineering capstone team at the University of California, Santa Barbara.
LC
Leo Che
LiDAR Sensor Integration
SENSINGJP
Johnny Perez
mmWave Radar Integration
SENSINGMH
Matthew Hsieh
Main Computer & ROS 2
COMPUTINGAG
Alex Gherghescu
Sensor Fusion & Navigation
ALGORITHMSCL
Chenghang (Peter) Liu
STM32, Motor Control & Hardware
INTEGRATIONCOURSEUCSB ECE 189A · Capstone Design
INSTRUCTORYogananda Isukapalli
TEACHING ASSISTANTAri Jacobs
FOLLOW THE BUILD
From concept
to clearer navigation.
Follow our development as we bring resilient sensing and autonomous navigation together.