Martijn Van Wezel Email & Phone Number

Netherlands

Utrecht, Nederland

Projects like:- PCB design to meet customer specifications.- Developed a time-accounting tool.- Transformed MS Visual Basic + Excel calibration process, resolving the four unknown value problem from - infinite time to microseconds.- Introduced an innovative water meter reader approach utilizing an affordable sensors and Wi-Fi chip for precise and cost-effective water usage monitoring, leveraging auto-correlation technique. (http://watermeter.muino.nl)

Delft En Omgeving, Nederland

The graduation was assessed with an 8.0A robust modular spiking neural networks training methodology for time-series datasets.With a focus on radar <> classification.Provisional U.S. patent: US 63/121.326

Pijnacker

Embedded engineer maintaining/developing servers, developing hardware (PCBs,), and software (embedded c/c++ applications and web applications (for internal management of the devices)). Working on a European funded project. Contact person for other companies. Worked with many types of wireless mesh networks/architectures and configured them.

Weteringweg 2, Pijnacker

The graduation was assessed with a 8/10.The goal was to build a tester for the wireless switching module (WSM). The WSM is used by the smart light industry to switch the light remotely on and off. There are several functions added to the WSM such as power usage measurement. SOWNet has to deliver around 2,500 modules at the end of summer 2016. To ensure the quality and good working of the WSMs, they should be tested and calibrated. The graduation project assignment is to develop a test tool that would calibrate, test and program the WSMs. First step in the process is to correctly define the requirements. Then the method of approach is to design the system in a high level and then divide it in subsystems according to their functionality. After that I developed four PCBs; one for each subsystem. The advantage of this approach is when a part of the design fails only this part needs to be reordered instead of the whole design. The test tool enclosure has no specific requirements and therefore its design comes at the last step. When the components and PCBs of one subsystem are received then the components are placed manually, component by component, to lower the risk of design errors.Finally the system was divided in four different PCBs: the test tool processor board, the 1000mA- reference power supply, the 30V reference power supply and the bed of needles.The test tool processor board was ordered first so that the software development could be started. After this board, the 1000mA PCB was ordered and delivered and is working as intended. The ordering of the other two boards is waiting for the test results. The software is also written for the test tool processor. The code of the waiting boards is written, but not whole software was implemented yet. The enclosure is also designed.

Delft En Omgeving, Nederland

During my internship at S[&]T vision I have investigated if the structure sensor can be used for 3D-measurements. Also there is a special interest in the measurement of staircases. There are new developments in structured light measurement devices and these needs to been reviewed and tested. Concerning visual odometry, there are mainly two concepts. The RGB odometry and the iterative closest point (ICP). The RGB odometry is an algorithm that uses features like SIFT points in the images. This was not interesting to be researched, because the structure sensor doesn’t have an RGB sensor (only an IR sensor). The IR sensor can only see the IR light and has no good visual of the environment. The ICP only needs point clouds. From these point clouds normal vectors needs to be calculated for the alignment of the 2 points clouds. The algorithm should be fast enough such that it can be used for real-time purposes, for example that an application can real-time measure and visualize the environment. There are several libraries that uses ICP and have an implementation of it. The implementation of the various libraries couldn’t be used directly and there have been some struggles implementing with the integration. The implementation of the Point Cloud Library (PCL) for the ICP algorithm was mainly for 3D printing of objects. There is a trunk version of ICP that I have transformed in my own implementation of ICP. The ICP algorithm works, but there are some troubles with alignment of point clouds. In my recommendation a concept is proposed, which would solve the problem. For the ICP algorithm, I have written a wrapper for C#, so the C++ libraries and my code can be used in C#.The degree of the internship = 8.8 out of 10.

Delft

vehicle recognition on the section control. The assignment was using machine learning approaches to recognize license plates and detection of trailers. The main methods that were used: Support Vector Machine, Viola Jones. I have completed the assignment successfully. The grade was for the report a 7.7 and for the internship a 9 out of 10.

Master Of Science (Msc.) And Ingenieur (Ir.), Computer Engineering Master

Finished in 1.5 years a 2 years study.

Schakelopleiding - Computer Engineering (Master)

Ingenieur (Ir.), Mechatronica, 8