Member Avatar for naureen

hi guys,

i am in final year of my bacelor degree. Its not a simple BCS honours that i am doing. Its infact computer engineering. It means i have studied robotics and now we are studing automation. SO Just 1 week remaining in choosing a project idea. Still i am looking for something new , innovative idea. I am willing to take an extreme and complex project.

My programming skills are damn good(c++ , c# , java). In my group i have 3 other persons. They are good at robtoics , databaes , and software engineering.

So any new , innovative and complex idea u guys can suggest will be more then welcome :)

My preference will be to take a project which has all elements. Like some Artifical Intellegence , Computer Vision , Robotics , Extreme Coding or even bioinformatics .

Thanks in advance :)

  • 2 Contributors
  • 3 Replies
  • 114 Views
  • 1 Day Discussion Span
  • Latest Post Latest Post by Ancient Dragon

Dani AI

Generated

— picking an "extreme" final project pays off if you treat it like a product: pick one clear real-world task, define measurable success criteria, and build the full pipeline (perception, decision, actuation, data, and UI). That lets each teammate contribute: the robotics specialist on hardware/control, the database person on logging and state, the software engineer on architecture and CI, and you on the core algorithms and integration.

Three compact but ambitious directions that combine AI, vision and robotics:

  • a mobile manipulator that locates, classifies and safely picks irregular objects in clutter using vision + learned grasping and classical motion planning;
  • a multi-robot inspection system that performs distributed mapping, detects visual anomalies with deep models, and schedules which robot handles each task;
  • an automated sample-handling pipeline (imaging + CV for detection/segmentation, ML for classification, robotics for precise handling) that includes a searchable experiment log and recovery routines.

To make this deliverable: define an MVP, then iterate. Suggested milestones: 1) narrow the use case and success metrics; 2) prototype perception in simulation or on prerecorded video using pre-trained models; 3) implement basic motion/control in simulation; 4) integrate perception and control in hardware-in-the-loop; 5) robustness testing and a short demo script. Split work by subsystem and keep well-defined interfaces (topics/messages, DB schemas, API stubs) so parallel work is possible.

Practical tips: start in simulation (reduces hardware risk), use transfer learning to cut labeling time, add extensive logging and replay tools, measure latency and failure modes, and plan a graceful fallback for the demo. As pointed out, simulating real-world throughput and timing early will save a lot of last-minute panic.

Recommended Answers

All 3 Replies

Member Avatar for Ancient Dragon

I don't know how much equipment (hardware) you have at your disposal, but one real-world example is:

Lets say we are in a factory that manufactures several different kinds of candy. After the candy is put in shipping containers (corrugated boxes) they roll down an assembly line past a barcode printer which prints a barcode that describes what is in the box (product code). At this point each kind of candy has its own conveyer belt so that all the boxes on a conveyer belt are the same thing. After that, the boxes flow into a single line, which might look like a Y shape. The boxes travel along the line past a barcode reader, at the rate of 1 box every 1/2 second, which reads the barcode and transmits it to a PC running your computer program. The computer program listens on the serial port for input from the barcode reader, when the data arrives the program uses the barcode data (I2OF5 format) and looks up a record in the database. After reading the database information, it formats the data and transmits it via another serial port or USB port to a printer that will print information on the sides of the box (like this one for example).

Your computer program must do all that in less than 1/2 second in order to keep up with the production line speed. The information to be printed on the box (which is in the database) is the product name, date manufactured, expiration date, lot number (for recall purposes) and up to four lines of product ingrediants.

Member Avatar for naureen

awww thanks so much.

just a question. Is that project really tough and challenging ?

PS, why we have to do all this in 1/2 second. As u said we must keep up the pace with production line. Is there already some production line standard like 1 second ?? So we can know that we will be improving prodcution line speed !

Member Avatar for Ancient Dragon

yes, I think it would be challenging for college students. You can simulate some of this by using a laptop or other computer to simulate the scanner (have it transmit barcode data to main computer every 1/2 second) and another computer or laptop to simulate a printer.

>>Is there already some production line standard like 1 second
the standard is 1/2 second, and management would like to make it faster than that!

There is actually more to the program than I revealed. For example, there is a switching center further down the production line that moves the boxes to different trucks for shipping to customers. The program also has a list of customers, the trucks and how many of each type candy do to each customer. The program will also have to send data to the switching center to tell it what box goes on what truck (robotics!). This production schedule is reset at the start of each shift (3 shifts/day, 24 hours a day, 7 days a week).

Be a part of the DaniWeb community

We're a friendly, industry-focused community of developers, IT pros, digital marketers, and technology enthusiasts meeting, networking, learning, and sharing knowledge.