Computer Vision Robotics Projects for Final Year Students (2026)
Computer vision gives robots the ability to detect objects, estimate pose, build maps, segment free space and close visual feedback loops. Combined with ROS 2, OpenCV and modern detectors, it enables navigation, manipulation and human–robot interaction.
This page lists 80+ high-impact CV-for-robotics topics. Tools include OpenCV, YOLO, MediaPipe, Detectron2, PyTorch, ROS 2 vision stacks and RealSense. Datasets span COCO, TUM RGB-D, KITTI, YCB, GraspNet and custom collections. Ideal for BE, BTech, MTech students in Bangalore and across India.
Core Frameworks & Tools
Libraries and sensors commonly used in computer vision robotics projects.
OpenCV
YOLO / Ultralytics
MediaPipe
PyTorch / Detectron2
ROS 2 Vision
RealSense / Depth
Best Computer Vision Robotics Topics, Tools & Datasets (80+)
Grouped by theme. Tools and typical datasets listed for each topic.
| # | Project Topic | Tools · Datasets |
|---|---|---|
| 🔍 Object Detection & Tracking for Robots | ||
| 1 | DetYOLO Object Detection for Mobile Robot Perception | YOLOv8/v11, COCO / custom, ROS 2 |
| 2 | DetReal-Time Detection on Edge (Jetson / RPi) | YOLO TensorRT, COCO subset |
| 3 | DetMulti-Object Tracking (SORT / DeepSORT / ByteTrack) | YOLO + tracker, MOT datasets |
| 4 | DetPerson Detection and Following Behavior | YOLO/MediaPipe, custom follow |
| 5 | DetTraffic Cone / Lane Marker Detection for AGV | YOLO, custom labeled set |
| 6 | DetWarehouse Pallet / Box Detection | YOLO, custom warehouse images |
| 7 | DetPPE (Helmet / Vest) Detection for Safety Robot | YOLO, PPE datasets |
| 8 | DetSmall Object Detection for Inspection Robots | YOLO fine-tune, custom defects |
| 9 | DetOpen-Vocabulary Detection (Grounding DINO style) Demo | Open-vocab models, COCO |
| 10 | DetDetection + ROS 2 Costmap Obstacle Layer | YOLO, Nav2 costmap plugin |
| 🗺️ Visual SLAM · Odometry · Mapping | ||
| 11 | SLAMORB-SLAM3 Monocular / Stereo / RGB-D Pipeline | ORB-SLAM3, TUM RGB-D / EuRoC |
| 12 | SLAMRTAB-Map RGB-D SLAM for Indoor Robots | RTAB-Map, RealSense, TUM |
| 13 | SLAMVisual Odometry (feature / direct methods) | OpenCV, TUM / KITTI odometry |
| 14 | SLAMLoop Closure Detection and Pose Graph Optimization | DBoW2 / NetVLAD, g2o |
| 15 | SLAMSemantic SLAM (objects as landmarks) | YOLO + SLAM, custom scenes |
| 16 | SLAMStereo Visual SLAM for Outdoor Rover | Stereo cameras, KITTI |
| 17 | SLAMIMU-Visual Fusion (VIO) for Drones / Robots | VINS-Fusion / OKVIS, EuRoC |
| 18 | SLAMMap Quality Metrics and Drift Analysis | TUM tools, ATE / RPE |
| 📐 Pose Estimation · Markers · 6D Pose | ||
| 19 | PoseArUco / AprilTag Pose Estimation for Robots | OpenCV ArUco, custom boards |
| 20 | Pose6D Object Pose Estimation (RGB / RGB-D) | PoseCNN / CosyPose, YCB-Video |
| 21 | PoseHand / Body Pose for HRI with MediaPipe | MediaPipe, custom gestures |
| 22 | PoseCamera–Robot Hand-Eye Calibration | OpenCV calibrateHandEye, chessboard |
| 23 | PoseMulti-Camera Extrinsic Calibration | OpenCV, kalibr-style |
| 24 | PoseCategory-Level Pose Estimation Demo | NOCS-style, synthetic + real |
| 25 | PosePose Tracking of Moving Objects for Grasp | Detector + pose refine, YCB |
| 🧩 Semantic / Instance Segmentation | ||
| 26 | SegFree-Space / Drivable Area Segmentation | DeepLab / YOLO-seg, Cityscapes |
| 27 | SegInstance Segmentation for Pick Targets | Mask R-CNN / YOLO-seg, COCO |
| 28 | SegSemantic Costmap from Segmentation | Seg model + Nav2 costmap |
| 29 | SegPlant / Crop Row Segmentation for Agri Robot | U-Net / YOLO-seg, agri datasets |
| 30 | SegDefect Segmentation for Inspection | U-Net, custom defect masks |
| 31 | SegPanoptic Segmentation for Indoor Scenes | Panoptic FPN, ADE20K / COCO |
| 32 | SegDepth-Aware Segmentation Fusion | RGB-D, RealSense, custom |
| ✋ Grasp Detection · Pick-and-Place Vision | ||
| 33 | GraspRectangle Grasp Detection (GG-CNN style) | GG-CNN / YOLO, Jacquard / Cornell |
| 34 | Grasp6-DoF Grasp Pose Detection | GraspNet / Contact-GraspNet, GraspNet-1B |
| 35 | GraspSuction Grasp Point Estimation | Depth CNN, custom suction set |
| 36 | GraspCluttered Bin Picking Vision Pipeline | Seg + grasp, YCB / custom bins |
| 37 | GraspTransparent / Reflective Object Handling | Specialized models, ClearGrasp-style |
| 38 | GraspGrasp Success Prediction from Vision | Classifier, robot trial data |
| 39 | GraspSim-to-Real Grasp Transfer Study | Isaac / Gazebo, domain randomization |
| 🎯 Visual Servoing · Tracking Control | ||
| 40 | ServoImage-Based Visual Servoing (IBVS) Demo | OpenCV, feature error, robot API |
| 41 | ServoPosition-Based Visual Servoing (PBVS) | Pose estimate + Cartesian control |
| 42 | ServoMarker-Based Docking / Precision Approach | ArUco, Nav2 docking concepts |
| 43 | ServoEye-in-Hand Tracking of Moving Target | Detector + visual servo loop |
| 44 | ServoHybrid Visual–Force Servoing Concept Demo | Vision + F/T sensor (sim or real) |
| 45 | ServoDirect Visual Servoing with Photometric Error | Research code, simple robot |
| 📏 Depth · Point Clouds · 3D Perception | ||
| 46 | 3DStereo Depth Estimation and Obstacle Map | OpenCV stereo, SGBM, KITTI |
| 47 | 3DMonocular Depth Estimation for Navigation | MiDaS / Depth Anything, custom |
| 48 | 3DPoint Cloud Clustering and Ground Removal | PCL / Open3D, RealSense |
| 49 | 3D3D Object Detection (PointPillars style) Demo | OpenPCDet, KITTI 3D |
| 50 | 3DRGB-D Scene Reconstruction for Manipulation | TSDF / KinectFusion-style, YCB |
| 51 | 3DLidar–Camera Calibration and Fusion | Calibration targets, KITTI |
| 🤝 HRI · Domain Robotics Vision | ||
| 52 | HRIGesture Recognition for Robot Commands | MediaPipe / YOLO, custom gestures |
| 53 | HRIFace / Emotion Cues for Social Robot | MediaPipe / FER models |
| 54 | HRIGaze / Attention Estimation for HRI | MediaPipe face mesh, custom |
| 55 | DomainAgricultural Fruit Detection and Counting | YOLO, MinneApple / custom |
| 56 | DomainRoad / Lane Detection for Autonomous Rover | Seg / detection, BDD100K / CULane |
| 57 | DomainMedical / Lab Sample Handling Vision | YOLO, custom lab objects |
| 58 | DomainConstruction Site Hazard Detection | YOLO, custom safety dataset |
| 59 | DomainRetail Shelf Monitoring from Mobile Robot | Detection + OCR optional |
| 60 | DomainUnderwater / Low-Visibility Enhancement + Detect | Enhancement + YOLO, underwater sets |
| 🔗 ROS 2 Vision Pipelines · Integration | ||
| 61 | ROSvision_msgs Detection Pipeline in ROS 2 | ROS 2, vision_msgs, YOLO node |
| 62 | ROSimage_pipeline: Rectify, Stereo, Depth | image_proc, stereo_image_proc |
| 63 | ROSCamera Calibration and Camera Info Management | camera_calibration, YAML |
| 64 | ROSTF2 Vision Frames and Optical Frame Conventions | tf2, REP-103/105 |
| 65 | ROSMulti-Camera Node and Sync Strategies | ROS 2, approximate time sync |
| 66 | ROSPerception → Nav2 / MoveIt Integration Demo | Full stack: detect → plan → act |
| 🔬 Advanced · Learning · Research-Oriented | ||
| 67 | AdvActive Vision: Next-Best-View for Reconstruction | View planning, RGB-D, sim |
| 68 | AdvDomain Adaptation for Robot Vision (sim→real) | DA methods, synthetic + real |
| 69 | AdvFew-Shot Object Detection for New SKUs | Few-shot detectors, custom |
| 70 | AdvSelf-Supervised / Contrastive Pretraining for Robots | SimCLR-style, robot images |
| 71 | AdvUncertainty Estimation in Detection for Safety | MC dropout / ensembles, COCO |
| 72 | AdvEvent Camera / Neuromorphic Vision Demo | DVS tools, event datasets |
| 73 | AdvNeural Radiance Fields Lite for Robot Scenes | NeRF / Instant-NGP, custom views |
| 74 | AdvVision-Language Models for Robot Instructions | CLIP / VLM, language goals |
| 75 | AdvAdversarial Robustness of Robot Detectors | Attacks / defenses, COCO |
| 76 | AdvMulti-Modal Fusion: Vision + Force / Tactile | RGB-D + tactile, custom |
| 77 | AdvContinual Learning for Changing Environments | CL methods, sequential tasks |
| 78 | AdvBenchmark: Detector Latency vs Accuracy on Robot HW | YOLO variants, Jetson / RPi |
| 79 | AdvDataset Creation Pipeline: Capture → Label → Train | CVAT / Label Studio, YOLO |
| 80 | AdvEnd-to-End Vision Robot: Detect → Localize → Grasp → Place | Full pipeline, YCB / custom |
| 81 | AdvSynthetic Data Generation for Robot Vision | Blender / Isaac, domain rand. |
| 82 | AdvCalibration-Free / Online Recalibration Strategies | Self-calibration, markers |
Topics reflect common robotics vision practice with open models and standard datasets. Contact us for training notes, ROS 2 nodes, evaluation metrics, university-format report, PPT and viva Q&A for any topic above.
Why Choose Us for CV Robotics Projects?
Bangalore-based guidance for BE, BTech and MTech students working on vision-enabled robots.
Detection & Tracking
YOLO pipelines, multi-object tracking and ROS 2 integration for mobile and industrial robots.
Visual SLAM
ORB-SLAM3, RTAB-Map and VIO with TUM/EuRoC evaluation and map quality analysis.
Pose & Grasp
6D pose, markers, grasp detection and pick-and-place vision with YCB and GraspNet-style data.
Servoing & HRI
Image-based visual servoing, docking and gesture/face pipelines for interactive robots.
Frequently Asked Questions — CV Robotics Projects
Top topics include YOLO detection for robots, visual SLAM (ORB-SLAM3 / RTAB-Map), 6D pose estimation, semantic free-space segmentation, grasp detection, visual servoing, person following and full detect→localize→grasp pipelines.
OpenCV, YOLO (ultralytics), MediaPipe, Detectron2, PyTorch, ROS 2 vision packages, RealSense; datasets include COCO, TUM RGB-D, KITTI, YCB, GraspNet, Cornell/Jacquard and custom collections.
Yes. Packages include model training/inference notes, ROS 2 nodes or Python pipelines, dataset guidance, evaluation metrics, university-format report, PPT and viva Q&A.
Vision provides perception (detect, segment, estimate pose) which feeds planners and controllers — detected objects become goals for Nav2 or MoveIt, depth maps update costmaps, and visual error drives servoing loops.
CV Robotics Lab — Bangalore
Cameras, depth sensors, GPU workstations and robot platforms for BE, BTech and MTech vision projects.
YOLO Detection
Robot Pipelines
Object DetectionRobot Pipelines
ORB-SLAM3 /
RTAB-Map
Visual SLAMRTAB-Map
6D Pose &
Markers
Pose EstimationMarkers
Grasp Detection
YCB / Jacquard
Pick VisionYCB / Jacquard
Visual Servoing
Loops
IBVS / PBVSLoops
RealSense
RGB-D Setup
Depth PerceptionRGB-D Setup
ROS 2 Vision
Nodes
IntegrationNodes
Viva & Report
Preparation
DocumentationPreparation