Automation has the potential to improve the standard of care but is difficult to realize due to perceptual challenges, especially in soft-tissue surgery. Machine learning can provide solutions, but typically requires large amounts of training data, which is time-consuming to collect. Even with shared platforms, hardware differences can prevent effective sharing of data between institutions. This letter proposes a standardized simulation platform for training and testing algorithms to control surgical robotic systems, which is built upon an open-source simulator, the Asynchronous Multi-Body Framework (AMBF), to enable quick prototyping of different scenes. An illustrative example of a suturing task on a phantom is presented and has formed the basis of a challenge, released to the community. The top-level contribution is the open-source release of a dynamic simulation environment that enables realistic suturing on a phantom, but supporting contributions include its extendable architectural design and a series of algorithmic optimizations to achieve real-time control and collision detection, realistic behavior of the needle and suture, and generation of multi-modal ground-truth data, including labeled depth data. These capabilities enable simulation-based surgical training and support research in machine learning for surgical scene perception and autonomous action.