Mahendran,
I believe the current algorithm for Robo is intentionally deterministic to help facilitate predictable test results. For many cases, it would be confusing and even surprising to interpret completely different results on a test run between two APKs that are exactly the same. Similarly, it would be confusing to add randomness in a written unit test or instrumentation test. When a test passes, you should be confident that running it any number of times would also only result in passes.
The easiest way to get Robo to explore more behavior, without writing any lines of code, is to simply increase the duration of the Robo test (unless you are finding that it just repeats the same paths somehow). You can adjust that along with the maximum depth using advanced settings. The advanced settings are initially hidden in the web UI. Please read the section "Configuring Robo Test" on this page:
https://firebase.google.com/docs/test-lab/robo-ux-test
When an app gets complex and you need to test more paths that Robo isn't finding, that's a good time to start writing instrumentation tests to specifically invoke those paths. If you find that Robo discovers a lot of your app but misses a few key paths, you could write instrumentation tests for just those paths, and run both Robo and the instrumentation tests to get better coverage. Eventually, the goal would be to cover more of your app in instrumentation and depend less on Robo, as you find time to write more of those tests yourself.
Doug