This paper describes a project that proposes a strategy to address the weaknesses in students abilities for abstract thinking and generic problem-solving skills. We suggest focusing on problemsolving skills outside textual programing by developing settings where descriptions of algorithmic solutions can be performed by physical and concrete constructions. It is a response to the decreasing performance in state-wide admission indicators of the cohorts arriving to university. This suggests a new way to teach is required. The proposal here is similar to new teaching trends around the work to teach computer science concepts and algorithms by implementing theories of constructivism, situated learning [11,16], active learning [15] and collaborative learning. Thus, we provide educational tools and "situated-leaning" activities that concur with others in the belief that these promote learning within an authentic context. We also follow the path that "kinesthetic learning" participation in such activities promotes learning by doing. However, the distinctive aspect of our approach is that the focus is on developing problem-solving skills using as a vehicle the situation and challenges of some computer science concepts. That is, the objective is not to teach computer science concepts (although some of these are developed). We also provide explicit instruction on problem-solving. In this paper we describe the approach with one activity. The activity progresses from a concrete individual experience in an outdoors environment or open gym, to working with concrete building material and then with a virtual environment in the computer. Usage of the computer is delayed and the last stage is indeed a textual based programming language; however, the progression is for reinforcing the problem-solving skills and is their applicability to larger scales of the same problem