Some thoughts on how to improve Sarthi for the visually impaired, and also for people with mild forms of disorientation.I assume that the shape of the handle clearly indicates the direction of the bearer. 

The user story is this: 

As an elderly I want Sarthi to indicate the way to a predefined destination, or back to my home. I want to use the walking stick as the user interface to my mobile phone, much like young people would use a smartwatch. 

Here are some suggestions that you might to turn into requirements in the case that the user story is accepted:

- Sarthi should use a virtual or magnetic compass to detect the heading of the user

  Note: an accelerometer chip with 2 axis acceleration and 1 axis yaw rate sensor would help to make it accurate and reactive

- Sarthi should have haptic feedback (e.g. to signal "turn left", "turn right", or length or pulse encoded distance to destination)

- Sarthi should have at least one capacitive user input element

  Note: a tapping code for questions like "how far", or commands like "next destination" or "go home" might be OK

A thought on Arduino: I would remove that requirement. Arduino is OK as a prototoping environment, but it's all but impossible to create portable Arduino code that uses energy management features in a meaningful way.

Hey Thomas! Thanks for the feedback. I am trying to build one in a few weeks. I too dont think Arduino is a great choice to go production. I used to use MSP but its too costly in low volumes. Been trying to learn STM8, partially gave up .The documentation is not that cool. Saw your project and will probably try that in a while :)

Hi Rohit, my advice is to concentrate on functional and system integration aspects first :-)

Later on, the choice of a microcontroller often depends on non-functional constraints  (cost per unit, programming infrastructure/community, availability, power requirements, etc).
By the way, the ultrasonic sensors for cars most likely require a specialized circuit. I wouldn't be surprised if mass-produced parking aid ECUs use an ASIC. Maybe you should also look into using video, a radar sensor, LIDAR or even a combination of cheap sensors and fuse the information. There are some terrific SoCs waiting to be used for something else than mobile phones or action cams.

In my opinion STM8 is a µC platform with rather readable documentation but it takes time to learn any µC. It's all a bit easier if you have an interactive environment, and that's where Forth comes in. The abstraction Arduino offers comes at a cost. You'll see that once you realize that it's either 8 K of bad performing Arduino code, or 0.8 K of C code tightly coupled  to a µCs peripherals delivering outstanding performance. But then again, performance is in the non-functional domain ;-)

Are you considering it purely as a transmitter or a receiver as well? A lot of people keep their sticks near them all the time, am audio visual or vibration alert that could be linked to something like a doorbell might be a good idea.