In CPU scheduling the classical problem of real-time and time-sharing operating systems is to schedule a sequence of jobs; the only given information, in this case, is the processing and completion time of the particular processes. All the previous scheduling algorithms focused on the system level abstraction for the scheduling decision functions or the functions which implement them. In this research, a new scheduling algorithm, an Adaptive Multi-level Queue scheduling, will maximise the CPU utilisation and cater to the energy efficiency of the embedded system. This scheduling algorithm can adapt to the environment. The scheduling algorithm will be helpful in the embedded devices which are using a fixed battery. The scheduler combines various types of well-established scheduling algorithms like – Round-robin, Priority (Pre-emptive) and Multi-Level Queue feedback algorithms. The Adaptive feature incorporates a sleeping function in the scheduling algorithm which helps in the system energy optimisation.