Browse Topic: Cameras

Items (595)
The empennage of a helicopter is largely responsible for its stability in forward flight. Its performance is mainly determined by its aerodynamics. In this paper, the empennage of a CoAX 2D ultralight research helicopter is analyzed in detail. For this purpose, the helicopter was equipped with flow measurement devices and flight tests were performed, covering different flight conditions. Measurements from a nose boom as well as the pilot’s control inputs and helicopter's position are available for evaluation. For the empennage in particular, seven-hole flow probes were mounted on it and various cameras were used to record the movement of the surface tufts.
Rottmann, LukasPaintner, RafaelMüller, DanielHunold, Julian
In this paper, we develop a new feature-based algorithm using stereo cameras to estimate stochastic ship-deck motion at high sea states. Unlike our previous algorithms, this algorithm is able to estimate the motion of arbitrary ship structures without prior information on the ship's visual appearance or geometry. The algorithm requires an initial pose and suffers from drift over time, which was resolved by fusing it with our previous 2D feature-based vision algorithm. The combined vision algorithm is validated using a simulated ship featuring 3D ship structures and 2D flight deck markings representative of a DDG-51 ship. The results indicate that the algorithm can accurately estimate the pose of a simulated ship undergoing Sea-State 6 motion. The vision algorithm was further validated in a simple free-flight test.
Chopra, Inderjit
The Shake-The-Box technique was applied to experimentally quantify the time-resolved volumetric flow field around a free-flying quadcopter UAV with an overall span of about 0.5 m. State-of-the-art LED illumination and high-speed camera equipment was combined with modern Lagrangian tracer particle tracking and data assimilation techniques, facilitating a measurement volume larger than 1.5m3. The setup allowed for both hover and limited maneuvering of the quadcopter, while resolving even small details of the complex interactional aerodynamics. In hover out of ground effect, the four individual rotor wakes merged into a single jet within a few rotor radii below the rotor planes. Evaluating the mass and momentum fluxes over suitable control volumes yields accurate estimates for the quadcopter's total thrust, the asymmetric thrust distribution between front and back rotors, and the entrainment of external flow through turbulent mixing. Hover in ground effect decreases the power requirement and induces recirculating flow in the center of the four rotors. The outwash pattern is non-uniform with jets developing between the rotors and pointing in radially outward directions. Forward flight cases result in a skewed, rapidly merging wake flanked by the roll-up of two "super-vortices" similar to the wingtip vortices of fixed-wing vehicles.
Wolf, Claus ChristianSchröder, AndreasStrübing, TobiasBosbach, JohannesHeintz, AlexanderSchwarz, ClemensSchanz, Daniel
The rotorcraft community faces significantly higher accident rates compared to fixed-wing commercial aircraft, underscoring the critical need for enhanced safety measures. While Helicopter Flight Data Monitoring programs hold promise in improving safety, their widespread adoption remains limited, partly due to challenges associated with the acquisition and analysis of flight data. This paper proposes a Deep Learning (DL) solution to address safety concerns within the rotorcraft community by efficiently acquiring and analyzing flight data for a more automated and comprehensive safety assessment. Specifically, we leverage data obtained with cost-effective off-the-shelf cameras, and process it through Convolutional Neural Networks for automated detection and classification of gauges from several helicopters' cockpits. Our DL pipeline integrates a classifier for helicopter identification, an object detector for cockpit gauges detection and classification, and a network to infer the reading of each detected gauge. The contribution of this work is two-fold: (1) enhance rotorcraft safety by developing a DL framework capable of detecting, classifying, and inferring gauge readings for different helicopter types, and (2) boost research in the field by constructing a curated dataset valuable for aviation and machine learning communities.
Khelifi, AmineJohnson, Charles C.Thompson, LaceyBouaynaya, Nidhal C.Carannante, GiuseppinaTrabelsi, Mohamed Ali
Object Detection and Tracking for Autonomous Vehicles in Adverse Weather Conditions2021-01-00794/6/2021
Object detection and tracking is a central aspect of perception for autonomous vehicles. While there has been significant development in this field in recent years, many perception algorithms still struggle to provide reliable information in challenging weather conditions which include night-time, direct sunlight, glare, fog, etc. To achieve full autonomy, there is a need for a robust perception system capable of handling such challenging conditions. In this paper, we attempt to bridge this gap by proposing an algorithm that combines the strength of automotive radars and infra-red thermal cameras. We show that these sensors complement each other well and provide reliable data in poor visibility conditions. We demonstrate the advantages of a thermal camera over a visible-range camera in these situations and employ YOLOv3 for object detection. The proposed system utilizes a modified Track-Oriented Multiple Hypothesis Tracking (MHT) algorithm which uses data from these sensors to keep track of the surrounding vehicles. The modifications in the well-known MHT algorithm were introduced in order to curb the exponential growth of possible hypotheses and consequently reduce the computational time without loss of any critical information. To validate the system, we provide a real-time implementation on an urban dataset collected at the Texas A&M University.
Bhadoriya, Abhay SinghVegamoor, Vamsi KrishnaRathinam, Sivakumar
Predicting Desired Temporal Waypoints from Camera and Route Planner Images using End-To-Mid Imitation Learning2021-01-00884/6/2021
This study is focused on exploring the possibilities of using camera and route planner images for autonomous driving in an end-to-mid learning fashion. The overall idea is to clone the humans’ driving behavior, in particular, their use of vision for ‘driving’ and map for ‘navigating’. The notion is that we humans use our vision to ‘drive’ and sometimes, we also use a map such as Google/Apple maps to find direction in order to ‘navigate’. We replicated this notion by using end-to-mid imitation learning. In particular, we imitated human driving behavior by using camera and route planner images for predicting the desired waypoints and by using a dedicated control to follow those predicted waypoints. Besides, this work also places emphasis on using minimal and cheaper sensors such as camera and basic map for autonomous driving rather than expensive sensors such Lidar or HD Maps as we humans do not use such sophisticated sensors for driving. Also, even after decades of research, the reasonable place for ‘mid’ in the End-to-End approach, as well as, the trade-off between data-driven and math-based approach is not fully understood. Therefore, we focused on the end-to-mid learning approach and tried to identify the reasonable place for ‘mid’ in the end-to-end pipeline.
Arul Doss, Aravind ChandradossGuvenc, Levent
Visual SLAM in Long-Range Autonomous Parking Application Based on Instance-Aware Semantic Segmentation via Multi-Task Network Cascades and Metric Learning Scheme2021-01-00774/6/2021
Long-range Autonomous Parking is becoming an attractive application in terms of demands. The vehicle is capable of driving autonomously into the appointed parking slot when the driver leaves it at the drop-off spot. In this application, the ability of accurate localization has become a key issue, especially in GPS-denied environments. This paper proposes a method of localization and mapping for Long-range Autonomous Parking, which is achieved by Visual SLAM based on deep learning algorithms. Firstly, we propose an instance segmentation via multi-task network cascades, and even in a complex visual environment, the main roadway instances of interest in the parking lot IPM image can be detected, such as parking corners, speed bumps. Then we combine the information of wheel encoders to build a global semantic map of the parking lot. Vehicles can often rely on semantic map matching to achieve high-precision localization. However, without a good initial position, it is difficult to infer an accurate position by matching the semantic map, such as randomly selecting entrances to enter the parking lot. Therefore, we propose an area feature network based on metric learning to extract features that distinguish different areas and infer the approximate initial position of the vehicle. Specifically, we extract features from the images of the surround-view cameras, use the vehicle position as weak supervision, and finally construct an area feature map. In summary, our proposed method provides accurate vehicle localization and parking lot maps for Long-range Autonomous Parking.
Yan, YixiongHang, YangHu, TianrenYu, HaoLai, Feng
Collins Aerospace, through its Common Avionics Architecture System (CAAS) and Flight2 avionics management systems for rotary wing aircrafts, provides extensive video processing, internal graphics generation, and overlay capabilities on real time video streamed from onboard EO/IR cameras providing situational awareness to the pilot in clear day-light and reduced visibility/night conditions. This capability has served our customers well in their cargo, assault, and multi-mission roles, improving the effectiveness of their missions. We now realize that more can be done to reduce pilot workload and enhance mission effectiveness by extracting visual intelligence from the video feed using machine vision. In this paper we explore the use of deep learning based computer vision to extract visual intelligence from onboard video feed and use it to automate low risk pilot actions, such as automatic detection and tracking of objects of interest, panning to maintain focus on objects, zooming on to a chosen object and providing contextualized data link message recommendations.
Huddar, VinayMudapaka, Nitesh
Autonomous Vehicles Camera Blinding Attack Detection Using Sequence Modelling and Predictive Analytics2020-01-07194/14/2020
Autonomous vehicles are waiting to address the global automotive mobility challenges through an intelligent smart transportation system, which includes advanced sensor-actuator configurations to control, navigate, and drive the vehicles. Multi-sensor data fusion from the key sensors such as camera, radar, and lidar is used to achieve the environmental perception for autonomous vehicles by capturing the various attributes of the environment. Cameras are the dominant sensors to achieve the perception by providing vision capability to vehicles. The direct interface of the cameras with the dynamic driving environment carries numerous attack surfaces on the camera. Blinding attacks on the cameras are one of the critical attacks with an intention to blind the cameras either fully or partially by projecting light into the cameras to hide the objects which results in failure in object detection. Here, the blinding attack detection approach is proposed which detects the blinding attacks on the camera in a dynamic driving environment by camera data predictive analytics. The proposed system predicts the future next frame of the video at each time instance and compares the received frame from the camera with the predicted frame at that instance to detect the blinding attacks. The incoming frames from the camera are sequentially modeled using a convolutional encoder-decoder neural network to predict the consecutive future frames, and the predicted frames are compared with the received camera frames to identify the similarity measure between the predicted and incoming camera frames of the same instance. Further, the approach detects the blinding on the camera, if the similarity measure calculated falls below a fixed threshold. The similarity measure which is inversely proportional to the amount of blinding is used to identify the blinding attacks. The predictive analytics of the sequentially modeled video frames with similarity measurement is used for the successful detection of blinding attacks.
D H, Sharath YadavAnsari, Asadullah
SUV Kinematics during a Steer-Induced Rollover Resolved Using Consumer-Grade Video, Laser Scans and Match-Moving Techniques2020-01-06424/14/2020
Rollover crashes are complex events that generate motions in all six degrees of freedom (6DOF). Directly quantifying the angular rotations from video can be difficult and vehicle orientation as a function of time is often not reported for staged rollover crashes. Our goal was to evaluate the ability of using a match-moving technique and consumer-grade video cameras to quantify the roll, pitch and yaw angles and angular velocities of a rollover crash. We staged a steer-induced rollover of an SUV at 106 km/h. The vehicle was fitted with tri-axial accelerometers and angular rate sensors, and five consumer-grade video cameras (2 on tripods, 2 on drones, 1 handheld, ~30 fps) captured the event. Roll, pitch and yaw angles were determined from the video using specialized software. We then compared the vehicle orientation angles from the video data to the integrated angular rate data measured by onboard sensors, and also compared the angular rates from the differentiated video data to the angular rates measured directly by the sensors. We found that both methods of measuring the 3D angles and angular rates generated similar results. The integrated sensor data drifted a maximum of 13° relative to the video-based angles, with RMS differences of ±2.7° or less when the drift was removed. The differentiated video data did not drift relative to the sensor data, with RMS differences of ±0.22 rad/s or less. These findings indicate that both methods generate similar results and are suitable for reconstructing rollovers. Given the drift we observed in the integrated sensor data, we recommend using angle measurements from the video to quantify the amount of drift in integrated sensor data if accurate knowledge of the vehicle’s orientation as a function of time is important.
Young, Cole R.King, David J.Siegmund, Gunter P.
Items per page:
1 – 50 of 595