H. Oliver Gao

Ronan Keane, H. Oliver Gao

Lane changing dynamics are an important part of traffic microsimulation and are vital for modeling weaving sections and merge bottlenecks. However, there is often much more emphasis placed on car following and gap acceptance models, whereas lane changing dynamics such as tactical, cooperation, and relaxation models receive comparatively little attention. This paper develops a general relaxation model which can be applied to an arbitrary parametric or nonparametric microsimulation model. The relaxation model modifies car following dynamics after a lane change, when vehicles can be far from equilibrium. Relaxation prevents car following models from reacting too strongly to the changes in space headway caused by lane changing, leading to more accurate and realistic simulated trajectories. We also show that relaxation is necessary for correctly simulating traffic breakdown with realistic values of capacity drop.

Ronan Keane, H. Oliver Gao

Many problems involve the use of models which learn probability distributions or incorporate randomness in some way. In such problems, because computing the true expected gradient may be intractable, a gradient estimator is used to update the model parameters. When the model parameters directly affect a probability distribution, the gradient estimator will involve score function terms. This paper studies baselines, a variance reduction technique for score functions. Motivated primarily by reinforcement learning, we derive for the first time an expression for the optimal state-dependent baseline, the baseline which results in a gradient estimator with minimum variance. Although we show that there exist examples where the optimal baseline may be arbitrarily better than a value function baseline, we find that the value function baseline usually performs similarly to an optimal baseline in terms of variance reduction. Moreover, the value function can also be used for bootstrapping estimators of the return, leading to additional variance reduction. Our results give new insight and justification for why value function baselines and the generalized advantage estimator (GAE) work well in practice.