Relay communication systems have attracted great attention recently due to their advantages in coverage extension, improved signal quality as well as increased end-to-end throughput. In the multiple-input multiple-output (MIMO) spatial-division multiplexing (SDM) amplify-and-forward (AF) relay communication systems, the bit error rate (BER) performance of the system depends significantly on the signal detector at the destination. However, optimal detectors which provide the minimum BER often require probitive complexity. In order to make it possible for practical implementation, we propose to apply the lattice reduction (LR) to the linear detectors at the destination receiver in order to balance the trade-off between their detection performance and computational complexity. Our analysis shows that the LR-aided (LRA) linear detectors using zero forcing (ZF) and minimum mean square error (MMSE) can achieve signifcant improvement in BER performance over the linear ZF and MMSE detectors while requiring the same complexity order.