WO2021025645A1

WO2021025645A1 - An apparatus for recognizing damages in a land vehicle and a method to operate the said device thereof

Info

Publication number: WO2021025645A1
Application number: PCT/TR2020/050241
Authority: WO
Inventors: Serhan GÖKÇEBAĞ; Pinar GÜRKAN
Original assignee: Jungotech İleri̇ Teknoloji̇ Çözümleri̇ Mühendi̇sli̇k Li̇mi̇ted Şi̇rketi̇
Priority date: 2020-03-26
Filing date: 2020-03-26
Publication date: 2021-02-11

Abstract

The invention discloses an fixed inspection system including a fixed apparatus. The sides, ceiling and door of the apparatus provides a calibrated environment that can admit a road vehicle inside wherein the sensors located take images under lights located inside on certain locations. A terminal is set for identification and damage data input either by the user or traded staff. The images taken and data gathered are either inspected by human intervention on cloud server or processed automatically to recognize the scope and extent of the damage on the vehicle if exists.

Description

AN APPARATUS FOR RECOGNIZING DAMAGES IN A LAND VEHICLE AND A METHOD TO OPERATE THE SAID DEVICE THEREOF

The invention disclosed is an apparatus and a computer implemented method to operate the said apparatus for fast, accurate and safe detection of damages on road vehicles. Primary technical field of the invention lies in detection of damages on road vehicles preventing paperwork, wait times, labor costs and fraud.

Damage detection and methods and apparatus for accurate claiming has quite a long history through the development of technology. As seen in our searches, we hereby denote some close documents but have no intention to cover all relevant documents. These documents are intended for better understanding of the invention with differences from some examples; some close some distant.

US7497618 discloses a mobile apparatus for damage detection. Being loaded to a truck, it includes a pivotable table to manage the inspection process. This has some drawbacks such as leveling, loading, calibration, radiation requirement and timing. Leveling is a major issue in mobile inspection as the truck stops at a random location. So, in order to obtain a safe detection, the first thing to be done is leveling, as stated in the said document. This is a hard task to succeed; it requires additional apparatus (for example additional leveling wheels in the said document) which of course does not guarantee a solid operation. Besides, these additional apparatus constitute an additional weight penalty for the vehicle to be inspected. Moreover, they generate a second problem; shall the truck (or trailer, etc.) be leveled before or after loading the vehicle to be inspected By its nature, a damaged vehicle may not be able to move by itself and it may not be loadable to an enclosure. Additional damages may occur, because lifting is a must in this case. Besides, there is the calibration problem for detection in various external conditions, so a calibration shall be done frequently, possibly before each and every detection which constitutes a big operational problem. The final drawback of the apparatus proposed in the said documents to be mentioned herein is the timing problem. The main problem aimed at the invention disclosed herein is timing problem. Timing problem refers to the wait time for appraisers and the paperwork they are entitled to do. The apparatus in the said document does not bring any ease to this problem, so does not solve the problem of rising costs of courtesy car nor late payments and repairs. The obvious difference in between this document and the invention disclosed is the method to operate the said apparatus; the said document does not disclose ways for automatic damage detection.

US9672440 discloses a damage recognition system by installing a camera on the vehicle. This is of course neither efficient, nor effective. It is not effective because one needs to install millions of cameras, if not billions. It has a huge cost beyond legal and ethical concerns that will occur during implementation. More to that, a camera or a pair of cameras as stated in the said document are simply incapable of detecting a damage. There are no pair (or even plurality) of cameras being able to see the whole outer side of the vehicle. So, the said document proposes an impossible way to recognize damages. Besides, it is not always favorable to recognize and assess damages automatically; in our searches we have been reported that users demand human testimony on images despite the advanced level of image recognition capabilities. This major concern of users can not be met by the said document.

There are other documents in the prior art such as damage estimation methods from images. They are not listed here due to absence of a standard in taking photos, so damage recognition and its evidence is practically improbable.

The invention disclosed is an automated or remotely controlled apparatus for damage detection. The road vehicle enters the apparatus and the apparatus detects the insurance-related damages. The current workflow for detecting damages is limited to the limited to the standard work hours and hence is prone to long waiting times from a few days to several weeks , which usually requires the damaged vehicle to hold. Beyond customer dissatisfaction, this increases losses of insurance companies via higher courtesy car rents. Besides the said problem, the invention disclosed solves the problem of fraud in damage claims. Recurrent, incorrect and fraudulent claims, a ratio by inclusion or misevaluation of appraisers lead to severe losses. Moreover, the invention disclosed aims reduction of the paperwork including time and labor lost for procedures. To solve all these, invention disclosed offers an instant, effective and secure way to detect and report damages on road vehicles.

The invention disclosed is an apparatus and method for damage recognition. The apparatus comprises a cabin wherein the dimensions of cabin are larger than that of the vehicles expected in order to allow the said vehicle to enter and be inspected by cameras, while leaving a reasonable margin for variances in the vehicle dimensions. Said dimensions shall be adapted with respect to the size of the aimed class of vehicles. As the majority of land vehicles are cars and motorcycles, we prefer a cabin size of 450x350 centimeters in length and width respectively. The cabin characteristics can be easily modified by a person skilled in the art, but for general purposes a height allowing the intake of damaged vehicles suffices. Obviously, this cabin can include a door, preferably on the short side. Based on our experimental and simulation work, we prefer a whole-side door opening outwards being connected from upside.

This cabin is opaque and lets no or little amount of light from outside for accurate calibrated photography of the suspected areas. Although this is critical for calibrated photo taking, the alternative of this is calibrating with respect to the lighting conditions of the apparatus as it has a fixed location. This fixed cabin has an array of cameras where these cameras are located at right angles to the plane on both axis pointing inwards of the cabin. This provides a metric representation of objects on the images taken. This is important for accurate processing of images, but not essential for the invention, as calibration and stretching of images as to form a parallelogram can easily be done by a person skilled in the art.

These cameras form a grid of sensors wherein spaced preferably equal, each aiming distinct parts of the vehicle. Therefore, there are a plurality of cameras aiming each a different part of the vehicle. For example, at least a camera points towards the right-front door, in a reasonable tolerance for possible variances in the car length or height. Car width also depends on the model, but invariant for scope of the camera as infinite focus cameras take precedence in the industry.

The invention disclosed also a terminal comprising a screen and means of connection for other devices. The said screen is a touch display for input and viewing of images. The alternative configuration for this is a traditional screen and a keyboard and a mouse, but the touch screen is preferable for the ease of use. This screen makes user interaction available where users can view the images taken by the camera grid, input their documents via the connectivity apparatus, enter their identification information (insurance type, company, insurance number, etc), the other stakeholders (other parties joining the incident, medical reports, incident report, police reports, etc.). As a result of intake of such information, the procedure eliminates additional human touch for incident identification. All the information regarding the damage can be taken from the user within the terminal interaction.

Obviously, the terminal includes a computer; a processor, a hard drive, a screen means of data input. Preferably, the means of data input is set to be Bluetooth connection and touch-screen, but other methods can be used as they are available at the state of the art. Besides, the terminal includes a connection port adjusted for intake of the images taken by the plurality of cameras within the cabin as stated previously. This may be a wireless connection port, an HDMI port, a mini-HDMI port, a VGA port, a USB port or similar ports that make image transfer available; or a plurality chosen from this set of ports. Besides, the terminal includes a camera and microphone for video chat via a remote human.

The method of execution of the invention disclosed is as follows: As the vehicle is located therein and the door is closed afterwards, the terminal is set to activate the cameras located in the cabin. As cameras take photos of the vehicle, the images are retrieved onto the terminal. These images provide a 360-degree view of the vehicle, where all images are pooled at the terminal. These images are sent to the server for further online usage. At the terminal, user is asked for identification information, via the touch screen and is invited to upload available documents via the connectivity apparatus. As these documents are associated with the images taken, they are uploaded to the server as well. Besides, the terminal allows the user to contact the insurance company via audio/video calling, if supported by the insurer.

At the server side, the images are processed for damage recognition. The most efficient way to handle this is image processing; comparing the image taken to the non-damaged image by image processing. The level of difference in between these two leads to an estimation of damage. In our studies, we conclude that this is the fastest and most accurate way of accessing the damage. But as insurers, users or repair contractors may desire to assess the level of image by themselves, the invention disclosed also provides an interface for manual image recognition by a remote expert. In this interface, the said remote expert connects the server via a remote computer, assesses the damaged parts and the level of damage in the part by visual inspection of the captured images. This is important for non-conformity issues in a multilateral operation. The third option is generating a model on a machine learning platform by using the human assessment data taken from the interface thereof and using the said model as improvised over time.

The personal data, insurance data, vehicle data are sent to the server and all sides as associated with the image set obtained by the grid of cameras herein. The said set of information may include payment information and a version of the invention disclosed includes forwarding payment information for payment purposes together a double approved estimation of the cost of the damage.

An embodiment of the invention further includes an interface that repair contractors can opt in and make their offers after having received the images photo and data. This provides accurate monetization of the damages directly by the repair contractors. As an alternative, the user interface on the terminal may display and print via a printer attached to the terminal information of the best-bidding repair contractors instantly or recieve them later by text message or e-mail as more contractors provide repair offers.

The fixed vehicle system disclosed herein inspects and recognizes the damage, if exists. In addition to the described embodiments, the invention disclosed may include lights inside the cabin directed to the vehicle to be inspected. These lights are standard lights preferably being white cold lights available.

The description of drawings are as follows;

Fig.1

denotes an example working principles of the invention disclosed.

Fig.2

denotes an example use of the invention disclosed. The said vehicle comes through (a), is taken in (b) and operation started (c) as shown.

Fig.3

shows a reverse angle of the invention disclosed. This is an outer image for clarification wherein the terminal, the grid and the door is shown.

Examples

PTL1:

NPL1:

Claims

The invention disclosed is a damage recognition apparatus comprising;
A cabin generated by at least one vertical and one horizontal element designated to generate walls of a rectangular prism or a cube wherein the dimensions of the cube are reasonably larger than that of the class of vehicles to be inspected, preferably 450*350 centimeters for cars.

The said cabin having opaque walls;

The said cabin having plurality of sensors placed as a grid aiming all inside;

A terminal for user data input.
The apparatus in claim 1 wherein the sensor positions are arranged according to the body parts of a vehicle.
The apparatus in claim 1 wherein the sensors being cameras,
The apparatus in claim 1 wherein the terminal comprises a touch screen, a processor, a memory element and connection device for information transmittal.
The apparatus in claim 1 wherein the terminal comprises a camera and microphone wherein the said camera and microphone are associated with the processor.
A method for damages recognition capable of working on at least one terminal, a grid of sensors and a server comprising;
Retrieving a set of input where the input includes at least vehicle type, make, model, year, owner and insurance.

A set of camera images being retrieved to the terminal; the set of images being sent to the server associated with the user input;

The set of camera images being compared to the previous images of the same type of vehicle; results of this comparison being saved onto the said server;

The set of images, comparison and results being transferred onto a second terminal for human verification; human verification results being sent to the server and stored therein;

The comparison results, verification, human verification results being fed to a machine learning platform to generate a model for damages recognition;
The said results, human verification or the verification being stored in the server as final result to be presented to parties.