What topic of the diploma should a microprocessor engineer choose?

Question

Electronics

Good afternoon. I’m writing my diploma soon, I can’t figure out what topic to do, so that it would be interesting and in which case it would not be embarrassing to show it. I don’t want to take standard topics, they offer options from the “smoke detector” or “burglar alarm” series.
I study in the direction of microprocessor technology, the diploma should ideally be for a microcontroller (suppose, ATMega 128). Of the important requirements - firstly, it should not be smart home systems and similar hackneyed topics, and secondly, the device must be such that a student can make it in a year.
The topic can be from any area, it is desirable that the result is also useful to someone. As an example, the driver's biometric control system, but this topic is already taken.
Help a student :)

Answer 1

In the list for expulsion of such students who cannot come up with a topic for a diploma.
I would muddy the auto-growing system of plants. Automatic watering depending on soil moisture, light control. The growers will thank you.

Answer 2

An interesting topic is acoustic trilateration. For example, in the interests of determining the position of the shooter. Here you have GPS measurements, and wireless communications between network nodes (LoRa, Zigbee), and an analog AF path, and DSP. But it can be done within a year.

Answer 3

There is such a device
limi.ru/product/voltammeter_vfm-3
Can you develop an analogue?
The device has a jamb - the high-voltage part is not exactly isolated from the low-voltage neutral wire.
The topic from the point of view of microprocessor circuitry is interesting. Because of this, the usb is slightly connected to the neutral, and the phase on the neutral is dangerous for the measurement operator.

Answer 4

this is what I understand the thesis will be
asilish ??
https://www.facebook.com/yuri.panchul/posts/101574...
The latter is useful for a student's thesis: a study of how cache parameters affect processor performance on certain algorithms with a certain memory latency.
In addition to MIPS microAptiv UP, there is also MIPS microAptiv UC - a smaller core without caches, with a predictable memory access latency via the SRAM-like protocol. Here, too, material for a student's thesis - make a harness for it, like MIPSfpga + is a harness for MIPS microAptiv UP.Having such a job in your resume, you can get a job not only in Wave and Microchip Technology, but also in ARM, Intel, Apple, AMD, ST Microelectronics, Texas Instruments. And in Russian companies - Baikal, ELVIS and others.
You can also do thesis by creating verification environments for the MIPS microAptiv UC/UP components (TLB MMU, caches, DSP unit, DSPRAM, the whole core) on SystemVerilog, UVM, Portable Stimulus. With the latter, it is even easier to find a job.
Here is another basis for the thesis - a comparison of the performance of DSP algorithms on a processor with a DSP extension versus a processor without a DSP extension, taking into account the cost of the extension (static and dynamic power consumption, core sizes)

Answer 5

I ’ll also throw
in an acoustic target - I read about the shooter from above =). In TSNIITOCHMASH, it seems, there are acoustic targets. You can try to reproduce for an air rifle. The point of impact is displayed on a laptop with a painted virtual target, compared with a paper one.
There are ballistic complexes based on photo frames, in addition to the position, the speed of the bullet is also determined.
PS By the way, I wonder how the first one will work on subsonic and whether it will work at all

Answer 6

Support for popular neural networks STM32H7 has recently appeared under STM32, it even
recognizes objects from the camera (processing the image itself)
air, scans the terrain with lidar
And add a neural network to this (almost useless, but interesting)
Before you have time to raise all the sensors, well, don’t care, you say that this is for further expansion and is not used now
You will gain experience in wiring multilayer boards with BGA and SDRAM boards and installation, there will be more than one revision)
Everything is not standard and not banal - it's difficult

Answer 7

I would write my own implementation of the Risc-V kernel. With chess and poets.