Basically you want to create a smart contract of services. Release money when work is done.
I see problem in this model that who will independently verify that lawn is mowed properly and payment should be released?
How can system stop service taker abusing the system?
Most of the time you wouldn't need an independent figure. Most transactions will go smoothly. What you need to watch out for is the 1% which go awry. This could be because they're looking to exit scam, they're new accounts, or just generally being unpleasant.
If there is an issue, all it takes is one side to open up a dispute. It will then be up to both sides to present their case as best they can. This will be publicly laid out for people to vote on. Voting won't be free or up to abuse. There will be a certain stake involved. If side A wins over side B, side A get the funds of side B. This will disincentivise risky votes (devil advocates or those taking their chances), botting, and just silly gamblers. It will be up to the people to publicly decide who is right.
I devised this system as an alternative to a PayPal dispute system. If you have any experience selling or buying online, they can be a real pain. They're opaque, vague, centralised. They don't let you in the know or to even fairly present your case. There is plenty of exploitation and even social engineering by illegitimate users to abuse this system. With my plan I hope to reduce that abuse substantially if not totally.