\abstract{The sensitivity with which the $\rm \gamma$ ray sources can be studied is inversely proportional to the angular resolution of an extensive air shower (EAS) array. Therefore it is important to determine the EAS arrival direction as accurately as possible. In this paper, we present the methods that were recently implemented by GRAPES-3 experiment to improve its angular resolution. In the GRAPES-3 experiment, consisting of an array of $\sim$ 400 scintillator detectors, the arrival direction of the shower is determined from the relative arrival times of particles at different detectors. The arrival times measured by each detector is initially corrected for the fixed arrival time caused by the cables connecting the individual detectors to their respective time-to-digital converter channels. A new method was developed based on
the random walk technique to measure these fixed arrival times with a smaller uncertainty.
A study based on simulations was also performed to verify the efficacy of this method. The arrival times were further corrected for the conical shape of the shower front, the slope of which had exhibited a strong dependence on the shower size and age. The correction for these dependencies led to an improvement in the angular resolution of the array by a factor of $\sim$ 2. The angular resolution of GRAPES-3 array obtained through array division methods is $\rm 0.8^{\circ}$ for energies $\rm E>5\,TeV$, which improves to $\rm0.3^{\circ}$ at $\rm E>100\,TeV$, and finally approaches $\rm0.2^{\circ}$ at $\rm E>500\,TeV$.}