Innovation contests have been an important tool used in product research and development for companies. In the innovation contest literature, most papers assume the homogenous innovation contest model or the all-pay auction model. In this paper, we consider the random trial contest model and study the optimal award scheme. We show that, in this contest model, risk types of contestants play important roles in the award scheme, and the results are independent of the probability density function of the random shock. These generalize the work in literature. In addition, the risk aversion coefficient will decide the allocation manner in a multiple-winner scheme, i.e., a concave allocation manner or a convex allocation manner is optimal.