Pranav Chougule
Research

Active Research


My research develops the sensing, reasoning, and autonomy layers that enable robots to conduct rigorous nondestructive evaluation of civil infrastructure with minimal human supervision.

RoboticsSensingAutonomy

Multimodal Robotic Sensing for Autonomous Infrastructure Evaluation

Design and deployment of mobile robotic platforms equipped with heterogeneous sensor suites for end-to-end autonomous nondestructive evaluation of civil infrastructure. The system orchestrates sensor scheduling, pose estimation, and data fusion without human intervention.

ROS 2Sensor fusionPath planningSLAM
UltrasoundSignal ProcessingMachine Learning

Ultrasonic Pulse Velocity (UPV) Testing & ML Interpretation

Development of data-driven pipelines for interpreting UPV waveforms acquired by robotic end-effectors. Integration of Florence-based Vision-Language Models to reason over signal morphology and generate structured NDE reports.

UPV hardwareVLM promptingFeature extractionAnomaly detection
AEStructural Health Monitoring

Acoustic Emission Monitoring

Real-time detection and source localisation of acoustic emission events during active loading scenarios. The goal is to map damage evolution in structural elements using piezoelectric sensor arrays coupled with Bayesian source-location algorithms.

Piezo arraysWavelet analysisSource localisationBayesian inference
HyperspectralComputer Vision

Hyperspectral Imaging for Material Characterisation

Application of push-broom hyperspectral cameras mounted on robotic platforms for surface composition analysis. Spectral signatures enable non-contact detection of carbonation, moisture, and chemical deterioration in concrete and masonry.

Push-broom imagingSpectral unmixingPCA/ICAClassification
3D VisionGaussian SplattingReconstruction

Gaussian Splatting for 3D NDE Reconstruction

Use of 3D Gaussian Splatting to produce photorealistic, spatially-indexed reconstructions of inspected surfaces. These serve as persistent scene representations for registering multi-modal sensor data across inspection visits.

3DGSNeRF comparisonPoint cloud alignmentTexture mapping
LaserAM ConcreteSPIE 2026★ Featured

Laser-Based Surface Monitoring of Additively Manufactured Concrete

Laser displacement sensing for high-resolution monitoring of robotic CO₂ injection in fresh additively manufactured concrete. Accepted for presentation at SPIE Smart Structures + NDE, March 2026.

Laser triangulationCO₂ injectionAM ConcreteDisplacement mapping