Agent-based modelling allows us to study unobservable or unreproducible events as we can recreate them in a model and identify the responsible parameters. Presently we design an ABM of hominins gathering to study how foraging behaviour is determined by perception, individual decisions, and social interactions. In the model, a group of foragers is searching for resources and return with them back to a central-place. As recent great apes and members of hunter-gatherer societies learn extraordinary orientation skills and memorize knowledge about their environment, we implemented two different strategies. Using the first strategy the foragers perform searching trips in random directions, while the second enables them to target specific locations as they possess knowledge about resource occurrence. We used this model to study the gathering behaviour of Homo erectus in Java around 2 million years ago by recreating the Pleistocene environment in the area around the Sangiran fossil site. The environment offers different resources with varying energetic value, like plants or aquatic animals occurring in reconstructed river courses. In the model resource distribution is derived in the model from net primary productivity and is based upon present precipitation data, resulting in seasonal changes in the resource availability throughout a year. We analyzed the behaviour during the process of populating a new and unknown environment like an island, by removing the environmental knowledge from our foraging agents. Furthermore, we analyzed the behavioural patterns occurring during the process of accumulating knowledge over time by stepwise increasing the number of foragers with access to environmental knowledge. Our results show how gathering behaviour changes with varying environmental knowledge. We also gained new insights into in-group behaviour during the process of populating new areas as we observed interesting interactions and dynamics between the foragers.